+ sendMessage({ text })} />
+
+ );
+}
diff --git a/examples/next-openai/app/test-openai-file-search/page.tsx b/examples/next-openai/app/test-openai-file-search/page.tsx
index 266deec05497..777b67d1f28d 100644
--- a/examples/next-openai/app/test-openai-file-search/page.tsx
+++ b/examples/next-openai/app/test-openai-file-search/page.tsx
@@ -1,116 +1,36 @@
'use client';
+import ChatInput from '@/component/chat-input';
+import FileSearchView from '@/component/openai-file-search-view';
import { useChat } from '@ai-sdk/react';
import { DefaultChatTransport } from 'ai';
-import ChatInput from '@/component/chat-input';
-import { OpenAIFileSearchMessage } from '@/app/api/chat-openai-file-search/route';
+import { OpenAIFileSearchMessage } from '../api/chat-openai-file-search/route';
export default function TestOpenAIFileSearch() {
- const { error, status, sendMessage, messages, regenerate, stop } =
- useChatOpenAI Code Interpreter Test
+ + {messages.map(message => ( +
+ {message.role === 'user' ? 'User: ' : 'AI: '}
+ {message.parts.map((part, index) => {
+ switch (part.type) {
+ case 'text':
+ return
+ ))}
+
+ {part.text}
;
+ case 'tool-code_interpreter':
+ return
- ({
+ transport: new DefaultChatTransport({
+ api: '/api/chat-openai-image-generation',
+ }),
+ });
+
+ return (
+ >;
+}) {
+ switch (invocation.state) {
+ case 'output-available':
+ return (
+ >;
+}) {
+ switch (invocation.state) {
+ case 'input-available':
+ return (
+ >;
+}) {
+ switch (invocation.state) {
+ case 'input-available':
+ return (
+ (
embeddings: Array,
usage?: EmbeddingModelUsage,
response: Awaited<
- ReturnType['doEmbed']>
+ ReturnType['doEmbed']>
>['response'] = { headers: {}, body: {} },
providerMetadata?: Awaited<
- ReturnType['doEmbed']>
+ ReturnType['doEmbed']>
>['providerMetadata'],
-): EmbeddingModelV2['doEmbed'] {
+): EmbeddingModelV3['doEmbed'] {
return async ({ values }) => {
assert.deepStrictEqual(expectedValues, values);
return { embeddings, usage, response, providerMetadata };
diff --git a/packages/ai/src/embed/embed-many.ts b/packages/ai/src/embed/embed-many.ts
index 98cd420a63da..1fbe717a880e 100644
--- a/packages/ai/src/embed/embed-many.ts
+++ b/packages/ai/src/embed/embed-many.ts
@@ -1,4 +1,4 @@
-import { ProviderOptions } from '@ai-sdk/provider-utils';
+import { ProviderOptions, withUserAgentSuffix } from '@ai-sdk/provider-utils';
import { prepareRetries } from '../util/prepare-retries';
import { splitArray } from '../util/split-array';
import { UnsupportedModelVersionError } from '../error/unsupported-model-version-error';
@@ -11,6 +11,7 @@ import { TelemetrySettings } from '../telemetry/telemetry-settings';
import { Embedding, EmbeddingModel, ProviderMetadata } from '../types';
import { resolveEmbeddingModel } from '../model/resolve-model';
import { EmbedManyResult } from './embed-many-result';
+import { VERSION } from '../version';
/**
Embed several values using an embedding model. The type of the value is defined
@@ -92,10 +93,15 @@ Only applicable for HTTP-based providers.
abortSignal,
});
+ const headersWithUserAgent = withUserAgentSuffix(
+ headers ?? {},
+ `ai/${VERSION}`,
+ );
+
const baseTelemetryAttributes = getBaseTelemetryAttributes({
model,
telemetry,
- headers,
+ headers: headersWithUserAgent,
settings: { maxRetries },
});
@@ -148,7 +154,7 @@ Only applicable for HTTP-based providers.
const modelResponse = await model.doEmbed({
values,
abortSignal,
- headers,
+ headers: headersWithUserAgent,
providerOptions,
});
@@ -249,7 +255,7 @@ Only applicable for HTTP-based providers.
const modelResponse = await model.doEmbed({
values: chunk,
abortSignal,
- headers,
+ headers: headersWithUserAgent,
providerOptions,
});
diff --git a/packages/ai/src/embed/embed.test.ts b/packages/ai/src/embed/embed.test.ts
index 1138acb82f93..b6a3776bae3f 100644
--- a/packages/ai/src/embed/embed.test.ts
+++ b/packages/ai/src/embed/embed.test.ts
@@ -1,7 +1,7 @@
-import { EmbeddingModelV2 } from '@ai-sdk/provider';
+import { EmbeddingModelV3 } from '@ai-sdk/provider';
import assert from 'node:assert';
-import { beforeEach, describe, expect, it } from 'vitest';
-import { MockEmbeddingModelV2 } from '../test/mock-embedding-model-v2';
+import { beforeEach, describe, expect, it, vi } from 'vitest';
+import { MockEmbeddingModelV3 } from '../test/mock-embedding-model-v3';
import { MockTracer } from '../test/mock-tracer';
import { Embedding, EmbeddingModelUsage } from '../types';
import { embed } from './embed';
@@ -9,10 +9,16 @@ import { embed } from './embed';
const dummyEmbedding = [0.1, 0.2, 0.3];
const testValue = 'sunny day at the beach';
+vi.mock('../version', () => {
+ return {
+ VERSION: '0.0.0-test',
+ };
+});
+
describe('result.embedding', () => {
it('should generate embedding', async () => {
const result = await embed({
- model: new MockEmbeddingModelV2({
+ model: new MockEmbeddingModelV3({
doEmbed: mockEmbed([testValue], [dummyEmbedding]),
}),
value: testValue,
@@ -25,7 +31,7 @@ describe('result.embedding', () => {
describe('result.response', () => {
it('should include response in the result', async () => {
const result = await embed({
- model: new MockEmbeddingModelV2({
+ model: new MockEmbeddingModelV3({
doEmbed: mockEmbed([testValue], [dummyEmbedding], undefined, {
body: { foo: 'bar' },
headers: { foo: 'bar' },
@@ -50,7 +56,7 @@ describe('result.response', () => {
describe('result.value', () => {
it('should include value in the result', async () => {
const result = await embed({
- model: new MockEmbeddingModelV2({
+ model: new MockEmbeddingModelV3({
doEmbed: mockEmbed([testValue], [dummyEmbedding]),
}),
value: testValue,
@@ -63,7 +69,7 @@ describe('result.value', () => {
describe('result.usage', () => {
it('should include usage in the result', async () => {
const result = await embed({
- model: new MockEmbeddingModelV2({
+ model: new MockEmbeddingModelV3({
doEmbed: mockEmbed([testValue], [dummyEmbedding], { tokens: 10 }),
}),
value: testValue,
@@ -84,7 +90,7 @@ describe('result.providerMetadata', () => {
};
const result = await embed({
- model: new MockEmbeddingModelV2({
+ model: new MockEmbeddingModelV3({
doEmbed: mockEmbed(
[testValue],
[dummyEmbedding],
@@ -106,17 +112,20 @@ describe('result.providerMetadata', () => {
describe('options.headers', () => {
it('should set headers', async () => {
const result = await embed({
- model: new MockEmbeddingModelV2({
+ model: new MockEmbeddingModelV3({
doEmbed: async ({ headers }) => {
assert.deepStrictEqual(headers, {
'custom-request-header': 'request-header-value',
+ 'user-agent': 'ai/0.0.0-test',
});
return { embeddings: [dummyEmbedding] };
},
}),
value: testValue,
- headers: { 'custom-request-header': 'request-header-value' },
+ headers: {
+ 'custom-request-header': 'request-header-value',
+ },
});
assert.deepStrictEqual(result.embedding, dummyEmbedding);
@@ -126,7 +135,7 @@ describe('options.headers', () => {
describe('options.providerOptions', () => {
it('should pass provider options to model', async () => {
const result = await embed({
- model: new MockEmbeddingModelV2({
+ model: new MockEmbeddingModelV3({
doEmbed: async ({ providerOptions }) => {
expect(providerOptions).toStrictEqual({
aProvider: { someKey: 'someValue' },
@@ -154,7 +163,7 @@ describe('telemetry', () => {
it('should not record any telemetry data when not explicitly enabled', async () => {
await embed({
- model: new MockEmbeddingModelV2({
+ model: new MockEmbeddingModelV3({
doEmbed: mockEmbed([testValue], [dummyEmbedding]),
}),
value: testValue,
@@ -166,7 +175,7 @@ describe('telemetry', () => {
it('should record telemetry data when enabled', async () => {
await embed({
- model: new MockEmbeddingModelV2({
+ model: new MockEmbeddingModelV3({
doEmbed: mockEmbed([testValue], [dummyEmbedding], { tokens: 10 }),
}),
value: testValue,
@@ -186,7 +195,7 @@ describe('telemetry', () => {
it('should not record telemetry inputs / outputs when disabled', async () => {
await embed({
- model: new MockEmbeddingModelV2({
+ model: new MockEmbeddingModelV3({
doEmbed: mockEmbed([testValue], [dummyEmbedding], { tokens: 10 }),
}),
value: testValue,
@@ -207,12 +216,12 @@ function mockEmbed(
embeddings: Array,
usage?: EmbeddingModelUsage,
response: Awaited<
- ReturnType['doEmbed']>
+ ReturnType['doEmbed']>
>['response'] = { headers: {}, body: {} },
providerMetadata?: Awaited<
- ReturnType['doEmbed']>
+ ReturnType['doEmbed']>
>['providerMetadata'],
-): EmbeddingModelV2['doEmbed'] {
+): EmbeddingModelV3['doEmbed'] {
return async ({ values }) => {
assert.deepStrictEqual(expectedValues, values);
return { embeddings, usage, response, providerMetadata };
diff --git a/packages/ai/src/embed/embed.ts b/packages/ai/src/embed/embed.ts
index 7d547540475c..6c3027ad1d03 100644
--- a/packages/ai/src/embed/embed.ts
+++ b/packages/ai/src/embed/embed.ts
@@ -1,4 +1,4 @@
-import { ProviderOptions } from '@ai-sdk/provider-utils';
+import { ProviderOptions, withUserAgentSuffix } from '@ai-sdk/provider-utils';
import { resolveEmbeddingModel } from '../model/resolve-model';
import { assembleOperationName } from '../telemetry/assemble-operation-name';
import { getBaseTelemetryAttributes } from '../telemetry/get-base-telemetry-attributes';
@@ -9,6 +9,7 @@ import { TelemetrySettings } from '../telemetry/telemetry-settings';
import { EmbeddingModel } from '../types';
import { prepareRetries } from '../util/prepare-retries';
import { EmbedResult } from './embed-result';
+import { VERSION } from '../version';
/**
Embed a value using an embedding model. The type of the value is defined by the embedding model.
@@ -78,10 +79,15 @@ Only applicable for HTTP-based providers.
abortSignal,
});
+ const headersWithUserAgent = withUserAgentSuffix(
+ headers ?? {},
+ `ai/${VERSION}`,
+ );
+
const baseTelemetryAttributes = getBaseTelemetryAttributes({
model: model,
telemetry,
- headers,
+ headers: headersWithUserAgent,
settings: { maxRetries },
});
@@ -120,7 +126,7 @@ Only applicable for HTTP-based providers.
const modelResponse = await model.doEmbed({
values: [value],
abortSignal,
- headers,
+ headers: headersWithUserAgent,
providerOptions,
});
diff --git a/packages/ai/src/generate-image/generate-image.test.ts b/packages/ai/src/generate-image/generate-image.test.ts
index a08683a65483..3972b5bae25b 100644
--- a/packages/ai/src/generate-image/generate-image.test.ts
+++ b/packages/ai/src/generate-image/generate-image.test.ts
@@ -1,7 +1,7 @@
import {
- ImageModelV2,
- ImageModelV2CallWarning,
- ImageModelV2ProviderMetadata,
+ ImageModelV3,
+ ImageModelV3CallWarning,
+ ImageModelV3ProviderMetadata,
} from '@ai-sdk/provider';
import {
convertBase64ToUint8Array,
@@ -14,10 +14,11 @@ import {
expect,
it,
test,
+ vi,
vitest,
} from 'vitest';
import * as logWarningsModule from '../logger/log-warnings';
-import { MockImageModelV2 } from '../test/mock-image-model-v2';
+import { MockImageModelV3 } from '../test/mock-image-model-v3';
import { generateImage } from './generate-image';
const prompt = 'sunny day at the beach';
@@ -29,12 +30,18 @@ const jpegBase64 =
'/9j/4AAQSkZJRgABAQEAYABgAAD/2wBDAAgGBgcGBQgHBwcJCQgKDBQNDAsLDBkSEw8UHRofHh0aHBwgJC4nICIsIxwcKDcpLDAxNDQ0Hyc5PTgyPC4zNDL/2wBDAQkJCQwLDBgNDRgyIRwhMjIyMjIyMjIyMjIyMjIyMjIyMjIyMjIyMjIyMjIyMjIyMjIyMjIyMjIyMjIyMjIyMjL/wAARCAABAAEDASIAAhEBAxEB/8QAFQABAQAAAAAAAAAAAAAAAAAAAAb/xAAUEAEAAAAAAAAAAAAAAAAAAAAA/8QAFQEBAQAAAAAAAAAAAAAAAAAAAAX/xAAUEQEAAAAAAAAAAAAAAAAAAAAA/9oADAMBAAIRAxEAPwCdABmX/9k='; // 1x1 black JPEG
const gifBase64 = 'R0lGODlhAQABAIAAAAUEBAAAACwAAAAAAQABAAACAkQBADs='; // 1x1 transparent GIF
+vi.mock('../version', () => {
+ return {
+ VERSION: '0.0.0-test',
+ };
+});
+
const createMockResponse = (options: {
images: string[] | Uint8Array[];
- warnings?: ImageModelV2CallWarning[];
+ warnings?: ImageModelV3CallWarning[];
timestamp?: Date;
modelId?: string;
- providerMetaData?: ImageModelV2ProviderMetadata;
+ providerMetaData?: ImageModelV3ProviderMetadata;
headers?: Record;
}) => ({
images: options.images,
@@ -68,10 +75,10 @@ describe('generateImage', () => {
const abortController = new AbortController();
const abortSignal = abortController.signal;
- let capturedArgs!: Parameters[0];
+ let capturedArgs!: Parameters[0];
await generateImage({
- model: new MockImageModelV2({
+ model: new MockImageModelV3({
doGenerate: async args => {
capturedArgs = args;
return createMockResponse({
@@ -88,7 +95,9 @@ describe('generateImage', () => {
style: 'vivid',
},
},
- headers: { 'custom-request-header': 'request-header-value' },
+ headers: {
+ 'custom-request-header': 'request-header-value',
+ },
abortSignal,
});
@@ -99,14 +108,17 @@ describe('generateImage', () => {
aspectRatio: '16:9',
seed: 12345,
providerOptions: { 'mock-provider': { style: 'vivid' } },
- headers: { 'custom-request-header': 'request-header-value' },
+ headers: {
+ 'custom-request-header': 'request-header-value',
+ 'user-agent': 'ai/0.0.0-test',
+ },
abortSignal,
});
});
it('should return warnings', async () => {
const result = await generateImage({
- model: new MockImageModelV2({
+ model: new MockImageModelV3({
doGenerate: async () =>
createMockResponse({
images: [pngBase64],
@@ -130,7 +142,7 @@ describe('generateImage', () => {
});
it('should call logWarnings with the correct warnings', async () => {
- const expectedWarnings: ImageModelV2CallWarning[] = [
+ const expectedWarnings: ImageModelV3CallWarning[] = [
{
type: 'other',
message: 'Setting is not supported',
@@ -143,7 +155,7 @@ describe('generateImage', () => {
];
await generateImage({
- model: new MockImageModelV2({
+ model: new MockImageModelV3({
doGenerate: async () =>
createMockResponse({
images: [pngBase64],
@@ -158,11 +170,11 @@ describe('generateImage', () => {
});
it('should call logWarnings with aggregated warnings from multiple calls', async () => {
- const warning1: ImageModelV2CallWarning = {
+ const warning1: ImageModelV3CallWarning = {
type: 'other',
message: 'Warning from call 1',
};
- const warning2: ImageModelV2CallWarning = {
+ const warning2: ImageModelV3CallWarning = {
type: 'other',
message: 'Warning from call 2',
};
@@ -171,7 +183,7 @@ describe('generateImage', () => {
let callCount = 0;
await generateImage({
- model: new MockImageModelV2({
+ model: new MockImageModelV3({
maxImagesPerCall: 1,
doGenerate: async () => {
switch (callCount++) {
@@ -200,7 +212,7 @@ describe('generateImage', () => {
it('should call logWarnings with empty array when no warnings are present', async () => {
await generateImage({
- model: new MockImageModelV2({
+ model: new MockImageModelV3({
doGenerate: async () =>
createMockResponse({
images: [pngBase64],
@@ -217,7 +229,7 @@ describe('generateImage', () => {
describe('base64 image data', () => {
it('should return generated images with correct mime types', async () => {
const result = await generateImage({
- model: new MockImageModelV2({
+ model: new MockImageModelV3({
doGenerate: async () =>
createMockResponse({
images: [pngBase64, jpegBase64],
@@ -248,7 +260,7 @@ describe('generateImage', () => {
it('should return the first image with correct mime type', async () => {
const result = await generateImage({
- model: new MockImageModelV2({
+ model: new MockImageModelV3({
doGenerate: async () =>
createMockResponse({
images: [pngBase64, jpegBase64],
@@ -277,7 +289,7 @@ describe('generateImage', () => {
];
const result = await generateImage({
- model: new MockImageModelV2({
+ model: new MockImageModelV3({
doGenerate: async () =>
createMockResponse({
images: uint8ArrayImages,
@@ -311,7 +323,7 @@ describe('generateImage', () => {
let callCount = 0;
const result = await generateImage({
- model: new MockImageModelV2({
+ model: new MockImageModelV3({
maxImagesPerCall: 2,
doGenerate: async options => {
switch (callCount++) {
@@ -325,7 +337,10 @@ describe('generateImage', () => {
providerOptions: {
'mock-provider': { style: 'vivid' },
},
- headers: { 'custom-request-header': 'request-header-value' },
+ headers: {
+ 'custom-request-header': 'request-header-value',
+ 'user-agent': 'ai/0.0.0-test',
+ },
abortSignal: undefined,
});
return createMockResponse({
@@ -339,7 +354,10 @@ describe('generateImage', () => {
size: '1024x1024',
aspectRatio: '16:9',
providerOptions: { 'mock-provider': { style: 'vivid' } },
- headers: { 'custom-request-header': 'request-header-value' },
+ headers: {
+ 'custom-request-header': 'request-header-value',
+ 'user-agent': 'ai/0.0.0-test',
+ },
abortSignal: undefined,
});
return createMockResponse({
@@ -356,7 +374,9 @@ describe('generateImage', () => {
aspectRatio: '16:9',
seed: 12345,
providerOptions: { 'mock-provider': { style: 'vivid' } },
- headers: { 'custom-request-header': 'request-header-value' },
+ headers: {
+ 'custom-request-header': 'request-header-value',
+ },
});
expect(result.images.map(image => image.base64)).toStrictEqual(
@@ -370,7 +390,7 @@ describe('generateImage', () => {
let callCount = 0;
const result = await generateImage({
- model: new MockImageModelV2({
+ model: new MockImageModelV3({
maxImagesPerCall: 2,
doGenerate: async options => {
switch (callCount++) {
@@ -382,7 +402,10 @@ describe('generateImage', () => {
size: '1024x1024',
aspectRatio: '16:9',
providerOptions: { 'mock-provider': { style: 'vivid' } },
- headers: { 'custom-request-header': 'request-header-value' },
+ headers: {
+ 'custom-request-header': 'request-header-value',
+ 'user-agent': 'ai/0.0.0-test',
+ },
abortSignal: undefined,
});
return createMockResponse({
@@ -397,7 +420,10 @@ describe('generateImage', () => {
size: '1024x1024',
aspectRatio: '16:9',
providerOptions: { 'mock-provider': { style: 'vivid' } },
- headers: { 'custom-request-header': 'request-header-value' },
+ headers: {
+ 'custom-request-header': 'request-header-value',
+ 'user-agent': 'ai/0.0.0-test',
+ },
abortSignal: undefined,
});
return createMockResponse({
@@ -415,7 +441,9 @@ describe('generateImage', () => {
aspectRatio: '16:9',
seed: 12345,
providerOptions: { 'mock-provider': { style: 'vivid' } },
- headers: { 'custom-request-header': 'request-header-value' },
+ headers: {
+ 'custom-request-header': 'request-header-value',
+ },
});
expect(result.warnings).toStrictEqual([
@@ -436,7 +464,7 @@ describe('generateImage', () => {
const maxImagesPerCallMock = vitest.fn(maxImagesPerCall);
const result = await generateImage({
- model: new MockImageModelV2({
+ model: new MockImageModelV3({
maxImagesPerCall: maxImagesPerCallMock,
doGenerate: async options => {
switch (callCount++) {
@@ -452,6 +480,7 @@ describe('generateImage', () => {
},
headers: {
'custom-request-header': 'request-header-value',
+ 'user-agent': 'ai/0.0.0-test',
},
abortSignal: undefined,
});
@@ -468,6 +497,7 @@ describe('generateImage', () => {
providerOptions: { 'mock-provider': { style: 'vivid' } },
headers: {
'custom-request-header': 'request-header-value',
+ 'user-agent': 'ai/0.0.0-test',
},
abortSignal: undefined,
});
@@ -485,7 +515,9 @@ describe('generateImage', () => {
aspectRatio: '16:9',
seed: 12345,
providerOptions: { 'mock-provider': { style: 'vivid' } },
- headers: { 'custom-request-header': 'request-header-value' },
+ headers: {
+ 'custom-request-header': 'request-header-value',
+ },
});
expect(result.images.map(image => image.base64)).toStrictEqual(
@@ -503,7 +535,7 @@ describe('generateImage', () => {
it('should throw NoImageGeneratedError when no images are returned', async () => {
await expect(
generateImage({
- model: new MockImageModelV2({
+ model: new MockImageModelV3({
doGenerate: async () =>
createMockResponse({
images: [],
@@ -527,13 +559,14 @@ describe('generateImage', () => {
it('should include response headers in error when no images generated', async () => {
await expect(
generateImage({
- model: new MockImageModelV2({
+ model: new MockImageModelV3({
doGenerate: async () =>
createMockResponse({
images: [],
timestamp: testDate,
headers: {
'custom-response-header': 'response-header-value',
+ 'user-agent': 'ai/0.0.0-test',
},
}),
}),
@@ -548,6 +581,7 @@ describe('generateImage', () => {
modelId: expect.any(String),
headers: {
'custom-response-header': 'response-header-value',
+ 'user-agent': 'ai/0.0.0-test',
},
},
],
@@ -559,7 +593,7 @@ describe('generateImage', () => {
const testHeaders = { 'x-test': 'value' };
const result = await generateImage({
- model: new MockImageModelV2({
+ model: new MockImageModelV3({
doGenerate: async () =>
createMockResponse({
images: [pngBase64],
@@ -582,7 +616,7 @@ describe('generateImage', () => {
it('should return provider metadata', async () => {
const result = await generateImage({
- model: new MockImageModelV2({
+ model: new MockImageModelV3({
doGenerate: async () =>
createMockResponse({
images: [pngBase64, pngBase64],
diff --git a/packages/ai/src/generate-image/generate-image.ts b/packages/ai/src/generate-image/generate-image.ts
index 8f938b9b0e00..eda33302caa2 100644
--- a/packages/ai/src/generate-image/generate-image.ts
+++ b/packages/ai/src/generate-image/generate-image.ts
@@ -1,5 +1,5 @@
-import { ImageModelV2, ImageModelV2ProviderMetadata } from '@ai-sdk/provider';
-import { ProviderOptions } from '@ai-sdk/provider-utils';
+import { ImageModelV3, ImageModelV3ProviderMetadata } from '@ai-sdk/provider';
+import { ProviderOptions, withUserAgentSuffix } from '@ai-sdk/provider-utils';
import { NoImageGeneratedError } from '../error/no-image-generated-error';
import {
detectMediaType,
@@ -15,6 +15,7 @@ import { ImageGenerationWarning } from '../types/image-model';
import { ImageModelResponseMetadata } from '../types/image-model-response-metadata';
import { GenerateImageResult } from './generate-image-result';
import { logWarnings } from '../logger/log-warnings';
+import { VERSION } from '../version';
/**
Generates images using an image model.
@@ -49,7 +50,7 @@ export async function generateImage({
/**
The image model to use.
*/
- model: ImageModelV2;
+ model: ImageModelV3;
/**
The prompt that should be used to generate the image.
@@ -115,7 +116,7 @@ Only applicable for HTTP-based providers.
*/
headers?: Record;
}): Promise {
- if (model.specificationVersion !== 'v2') {
+ if (model.specificationVersion !== 'v3') {
throw new UnsupportedModelVersionError({
version: model.specificationVersion,
provider: model.provider,
@@ -123,6 +124,11 @@ Only applicable for HTTP-based providers.
});
}
+ const headersWithUserAgent = withUserAgentSuffix(
+ headers ?? {},
+ `ai/${VERSION}`,
+ );
+
const { retry } = prepareRetries({
maxRetries: maxRetriesArg,
abortSignal,
@@ -151,7 +157,7 @@ Only applicable for HTTP-based providers.
prompt,
n: callImageCount,
abortSignal,
- headers,
+ headers: headersWithUserAgent,
size,
aspectRatio,
seed,
@@ -165,7 +171,7 @@ Only applicable for HTTP-based providers.
const images: Array = [];
const warnings: Array = [];
const responses: Array = [];
- const providerMetadata: ImageModelV2ProviderMetadata = {};
+ const providerMetadata: ImageModelV3ProviderMetadata = {};
for (const result of results) {
images.push(
...result.images.map(
@@ -214,13 +220,13 @@ class DefaultGenerateImageResult implements GenerateImageResult {
readonly images: Array;
readonly warnings: Array;
readonly responses: Array;
- readonly providerMetadata: ImageModelV2ProviderMetadata;
+ readonly providerMetadata: ImageModelV3ProviderMetadata;
constructor(options: {
images: Array;
warnings: Array;
responses: Array;
- providerMetadata: ImageModelV2ProviderMetadata;
+ providerMetadata: ImageModelV3ProviderMetadata;
}) {
this.images = options.images;
this.warnings = options.warnings;
@@ -233,7 +239,7 @@ class DefaultGenerateImageResult implements GenerateImageResult {
}
}
-async function invokeModelMaxImagesPerCall(model: ImageModelV2) {
+async function invokeModelMaxImagesPerCall(model: ImageModelV3) {
const isFunction = model.maxImagesPerCall instanceof Function;
if (!isFunction) {
diff --git a/packages/ai/src/generate-object/__snapshots__/generate-object.test.ts.snap b/packages/ai/src/generate-object/__snapshots__/generate-object.test.ts.snap
index e5d90ec7e305..dcd7676a2e73 100644
--- a/packages/ai/src/generate-object/__snapshots__/generate-object.test.ts.snap
+++ b/packages/ai/src/generate-object/__snapshots__/generate-object.test.ts.snap
@@ -7,6 +7,7 @@ exports[`generateObject > telemetry > should not record telemetry inputs / outpu
"ai.model.id": "mock-model-id",
"ai.model.provider": "mock-provider",
"ai.operationId": "ai.generateObject",
+ "ai.request.headers.user-agent": "ai/0.0.0-test",
"ai.response.finishReason": "stop",
"ai.settings.maxRetries": 2,
"ai.settings.output": "object",
@@ -22,6 +23,7 @@ exports[`generateObject > telemetry > should not record telemetry inputs / outpu
"ai.model.id": "mock-model-id",
"ai.model.provider": "mock-provider",
"ai.operationId": "ai.generateObject.doGenerate",
+ "ai.request.headers.user-agent": "ai/0.0.0-test",
"ai.response.finishReason": "stop",
"ai.response.id": "test-id-from-model",
"ai.response.model": "test-response-model-id",
@@ -56,6 +58,7 @@ exports[`generateObject > telemetry > should record telemetry data when enabled
"ai.prompt": "{"prompt":"prompt"}",
"ai.request.headers.header1": "value1",
"ai.request.headers.header2": "value2",
+ "ai.request.headers.user-agent": "ai/0.0.0-test",
"ai.response.finishReason": "stop",
"ai.response.object": "{"content":"Hello, world!"}",
"ai.response.providerMetadata": "{"testProvider":{"testKey":"testValue"}}",
@@ -88,6 +91,7 @@ exports[`generateObject > telemetry > should record telemetry data when enabled
"ai.prompt.messages": "[{"role":"user","content":[{"type":"text","text":"prompt"}]}]",
"ai.request.headers.header1": "value1",
"ai.request.headers.header2": "value2",
+ "ai.request.headers.user-agent": "ai/0.0.0-test",
"ai.response.finishReason": "stop",
"ai.response.id": "test-id-from-model",
"ai.response.model": "test-response-model-id",
diff --git a/packages/ai/src/generate-object/generate-object.test.ts b/packages/ai/src/generate-object/generate-object.test.ts
index 4481fa566f9c..f995d76d0e6c 100644
--- a/packages/ai/src/generate-object/generate-object.test.ts
+++ b/packages/ai/src/generate-object/generate-object.test.ts
@@ -6,7 +6,15 @@ import {
import { jsonSchema } from '@ai-sdk/provider-utils';
import { convertReadableStreamToArray } from '@ai-sdk/provider-utils/test';
import assert, { fail } from 'node:assert';
-import { afterEach, beforeEach, describe, expect, it, vitest } from 'vitest';
+import {
+ afterEach,
+ beforeEach,
+ describe,
+ expect,
+ it,
+ vitest,
+ vi,
+} from 'vitest';
import { z } from 'zod/v4';
import { verifyNoObjectGeneratedError as originalVerifyNoObjectGeneratedError } from '../error/verify-no-object-generated-error';
import * as logWarningsModule from '../logger/log-warnings';
@@ -14,6 +22,12 @@ import { MockLanguageModelV2 } from '../test/mock-language-model-v2';
import { MockTracer } from '../test/mock-tracer';
import { generateObject } from './generate-object';
+vi.mock('../version', () => {
+ return {
+ VERSION: '0.0.0-test',
+ };
+});
+
const dummyResponseValues = {
finishReason: 'stop' as const,
usage: {
@@ -250,6 +264,7 @@ describe('generateObject', () => {
modelId: 'test-response-model-id',
headers: {
'custom-response-header': 'response-header-value',
+ 'user-agent': 'ai/0.0.0-test',
},
body: 'test body',
},
@@ -265,6 +280,7 @@ describe('generateObject', () => {
modelId: 'test-response-model-id',
headers: {
'custom-response-header': 'response-header-value',
+ 'user-agent': 'ai/0.0.0-test',
},
body: 'test body',
});
@@ -523,6 +539,7 @@ describe('generateObject', () => {
doGenerate: async ({ headers }) => {
expect(headers).toStrictEqual({
'custom-request-header': 'request-header-value',
+ 'user-agent': 'ai/0.0.0-test',
});
return {
diff --git a/packages/ai/src/generate-object/generate-object.ts b/packages/ai/src/generate-object/generate-object.ts
index 9a1e6925f5c2..4fc552aa0156 100644
--- a/packages/ai/src/generate-object/generate-object.ts
+++ b/packages/ai/src/generate-object/generate-object.ts
@@ -4,6 +4,7 @@ import {
InferSchema,
ProviderOptions,
Schema,
+ withUserAgentSuffix,
} from '@ai-sdk/provider-utils';
import * as z3 from 'zod/v3';
import * as z4 from 'zod/v4';
@@ -42,6 +43,7 @@ import { getOutputStrategy } from './output-strategy';
import { parseAndValidateObjectResultWithRepair } from './parse-and-validate-object-result';
import { RepairTextFunction } from './repair-text';
import { validateObjectGenerationInput } from './validate-object-generation-input';
+import { VERSION } from '../version';
const originalGenerateId = createIdGenerator({ prefix: 'aiobj', size: 24 });
@@ -262,10 +264,15 @@ Default and recommended: 'auto' (best mode for the model).
const callSettings = prepareCallSettings(settings);
+ const headersWithUserAgent = withUserAgentSuffix(
+ headers ?? {},
+ `ai/${VERSION}`,
+ );
+
const baseTelemetryAttributes = getBaseTelemetryAttributes({
model,
telemetry,
- headers,
+ headers: headersWithUserAgent,
settings: { ...callSettings, maxRetries },
});
@@ -358,7 +365,7 @@ Default and recommended: 'auto' (best mode for the model).
prompt: promptMessages,
providerOptions,
abortSignal,
- headers,
+ headers: headersWithUserAgent,
});
const responseData = {
diff --git a/packages/ai/src/generate-speech/generate-speech.test.ts b/packages/ai/src/generate-speech/generate-speech.test.ts
index bf781d25eeee..66812fb5db7c 100644
--- a/packages/ai/src/generate-speech/generate-speech.test.ts
+++ b/packages/ai/src/generate-speech/generate-speech.test.ts
@@ -3,7 +3,15 @@ import {
SpeechModelV2,
SpeechModelV2CallWarning,
} from '@ai-sdk/provider';
-import { afterEach, beforeEach, describe, expect, it, vitest } from 'vitest';
+import {
+ afterEach,
+ beforeEach,
+ describe,
+ expect,
+ it,
+ vitest,
+ vi,
+} from 'vitest';
import * as logWarningsModule from '../logger/log-warnings';
import { MockSpeechModelV2 } from '../test/mock-speech-model-v2';
import { generateSpeech } from './generate-speech';
@@ -21,6 +29,12 @@ const mockFile = new DefaultGeneratedAudioFile({
const sampleText = 'This is a sample text to convert to speech.';
+vi.mock('../version', () => {
+ return {
+ VERSION: '0.0.0-test',
+ };
+});
+
const createMockResponse = (options: {
audio: GeneratedAudioFile;
warnings?: SpeechModelV2CallWarning[];
@@ -69,14 +83,19 @@ describe('generateSpeech', () => {
}),
text: sampleText,
voice: 'test-voice',
- headers: { 'custom-request-header': 'request-header-value' },
+ headers: {
+ 'custom-request-header': 'request-header-value',
+ },
abortSignal,
});
expect(capturedArgs).toStrictEqual({
text: sampleText,
voice: 'test-voice',
- headers: { 'custom-request-header': 'request-header-value' },
+ headers: {
+ 'custom-request-header': 'request-header-value',
+ 'user-agent': 'ai/0.0.0-test',
+ },
abortSignal,
providerOptions: {},
outputFormat: undefined,
@@ -226,6 +245,7 @@ describe('generateSpeech', () => {
timestamp: testDate,
headers: {
'custom-response-header': 'response-header-value',
+ 'user-agent': 'ai/0.0.0-test',
},
}),
}),
@@ -240,6 +260,7 @@ describe('generateSpeech', () => {
modelId: expect.any(String),
headers: {
'custom-response-header': 'response-header-value',
+ 'user-agent': 'ai/0.0.0-test',
},
},
],
diff --git a/packages/ai/src/generate-speech/generate-speech.ts b/packages/ai/src/generate-speech/generate-speech.ts
index e65067ce21ac..ddcfbaaabc5b 100644
--- a/packages/ai/src/generate-speech/generate-speech.ts
+++ b/packages/ai/src/generate-speech/generate-speech.ts
@@ -1,5 +1,5 @@
import { JSONValue, SpeechModelV2 } from '@ai-sdk/provider';
-import { ProviderOptions } from '@ai-sdk/provider-utils';
+import { ProviderOptions, withUserAgentSuffix } from '@ai-sdk/provider-utils';
import { NoSpeechGeneratedError } from '../error/no-speech-generated-error';
import { UnsupportedModelVersionError } from '../error/unsupported-model-version-error';
import { logWarnings } from '../logger/log-warnings';
@@ -15,7 +15,7 @@ import {
DefaultGeneratedAudioFile,
GeneratedAudioFile,
} from './generated-audio-file';
-
+import { VERSION } from '../version';
/**
Generates speech audio using a speech model.
@@ -122,6 +122,11 @@ Only applicable for HTTP-based providers.
});
}
+ const headersWithUserAgent = withUserAgentSuffix(
+ headers ?? {},
+ `ai/${VERSION}`,
+ );
+
const { retry } = prepareRetries({
maxRetries: maxRetriesArg,
abortSignal,
@@ -136,7 +141,7 @@ Only applicable for HTTP-based providers.
speed,
language,
abortSignal,
- headers,
+ headers: headersWithUserAgent,
providerOptions,
}),
);
diff --git a/packages/ai/src/generate-text/__snapshots__/generate-text.test.ts.snap b/packages/ai/src/generate-text/__snapshots__/generate-text.test.ts.snap
index bc5598da45d0..6bcedcbfd064 100644
--- a/packages/ai/src/generate-text/__snapshots__/generate-text.test.ts.snap
+++ b/packages/ai/src/generate-text/__snapshots__/generate-text.test.ts.snap
@@ -1091,6 +1091,7 @@ exports[`generateText > telemetry > should not record telemetry inputs / outputs
"ai.model.id": "mock-model-id",
"ai.model.provider": "mock-provider",
"ai.operationId": "ai.generateText",
+ "ai.request.headers.user-agent": "ai/0.0.0-test",
"ai.response.finishReason": "stop",
"ai.settings.maxRetries": 2,
"ai.usage.completionTokens": 10,
@@ -1105,6 +1106,7 @@ exports[`generateText > telemetry > should not record telemetry inputs / outputs
"ai.model.id": "mock-model-id",
"ai.model.provider": "mock-provider",
"ai.operationId": "ai.generateText.doGenerate",
+ "ai.request.headers.user-agent": "ai/0.0.0-test",
"ai.response.finishReason": "stop",
"ai.response.id": "test-id",
"ai.response.model": "mock-model-id",
@@ -1149,6 +1151,7 @@ exports[`generateText > telemetry > should record telemetry data when enabled 1`
"ai.prompt": "{"prompt":"prompt"}",
"ai.request.headers.header1": "value1",
"ai.request.headers.header2": "value2",
+ "ai.request.headers.user-agent": "ai/0.0.0-test",
"ai.response.finishReason": "stop",
"ai.response.providerMetadata": "{"testProvider":{"testKey":"testValue"}}",
"ai.response.text": "Hello, world!",
@@ -1180,6 +1183,7 @@ exports[`generateText > telemetry > should record telemetry data when enabled 1`
"ai.prompt.messages": "[{"role":"user","content":[{"type":"text","text":"prompt"}]}]",
"ai.request.headers.header1": "value1",
"ai.request.headers.header2": "value2",
+ "ai.request.headers.user-agent": "ai/0.0.0-test",
"ai.response.finishReason": "stop",
"ai.response.id": "test-id-from-model",
"ai.response.model": "test-response-model-id",
diff --git a/packages/ai/src/generate-text/content-part.ts b/packages/ai/src/generate-text/content-part.ts
index 81509a200dc7..e1cfb5d14290 100644
--- a/packages/ai/src/generate-text/content-part.ts
+++ b/packages/ai/src/generate-text/content-part.ts
@@ -1,6 +1,7 @@
import { ProviderMetadata } from '../types';
import { Source } from '../types/language-model';
import { GeneratedFile } from './generated-file';
+import { ReasoningOutput } from './reasoning-output';
import { TypedToolCall } from './tool-call';
import { TypedToolError } from './tool-error';
import { TypedToolResult } from './tool-result';
@@ -8,7 +9,7 @@ import { ToolSet } from './tool-set';
export type ContentPart =
| { type: 'text'; text: string; providerMetadata?: ProviderMetadata }
- | { type: 'reasoning'; text: string; providerMetadata?: ProviderMetadata }
+ | ReasoningOutput
| ({ type: 'source' } & Source)
| { type: 'file'; file: GeneratedFile; providerMetadata?: ProviderMetadata } // different because of GeneratedFile object
| ({ type: 'tool-call' } & TypedToolCall & {
diff --git a/packages/ai/src/generate-text/generate-text-result.ts b/packages/ai/src/generate-text/generate-text-result.ts
index c9bac1a6c4ff..61ef021bf50f 100644
--- a/packages/ai/src/generate-text/generate-text-result.ts
+++ b/packages/ai/src/generate-text/generate-text-result.ts
@@ -1,4 +1,3 @@
-import { ReasoningPart } from '@ai-sdk/provider-utils';
import { CallWarning, FinishReason, ProviderMetadata } from '../types';
import { Source } from '../types/language-model';
import { LanguageModelRequestMetadata } from '../types/language-model-request-metadata';
@@ -6,6 +5,7 @@ import { LanguageModelResponseMetadata } from '../types/language-model-response-
import { LanguageModelUsage } from '../types/usage';
import { ContentPart } from './content-part';
import { GeneratedFile } from './generated-file';
+import { ReasoningOutput } from './reasoning-output';
import { ResponseMessage } from './response-message';
import { StepResult } from './step-result';
import { DynamicToolCall, StaticToolCall, TypedToolCall } from './tool-call';
@@ -34,7 +34,7 @@ The text that was generated in the last step.
/**
The full reasoning that the model has generated in the last step.
*/
- readonly reasoning: Array;
+ readonly reasoning: Array;
/**
The reasoning text that the model has generated in the last step. Can be undefined if the model
diff --git a/packages/ai/src/generate-text/generate-text.test.ts b/packages/ai/src/generate-text/generate-text.test.ts
index 502cb088e1ee..e0c6402da3b0 100644
--- a/packages/ai/src/generate-text/generate-text.test.ts
+++ b/packages/ai/src/generate-text/generate-text.test.ts
@@ -31,6 +31,12 @@ import { GenerateTextResult } from './generate-text-result';
import { StepResult } from './step-result';
import { stepCountIs } from './stop-condition';
+vi.mock('../version', () => {
+ return {
+ VERSION: '0.0.0-test',
+ };
+});
+
const dummyResponseValues = {
finishReason: 'stop' as const,
usage: {
@@ -1677,9 +1683,10 @@ describe('generateText', () => {
const result = await generateText({
model: new MockLanguageModelV2({
doGenerate: async ({ headers }) => {
- assert.deepStrictEqual(headers, {
- 'custom-request-header': 'request-header-value',
- });
+ assert.equal(
+ headers?.['custom-request-header'],
+ 'request-header-value',
+ );
return {
...dummyResponseValues,
@@ -1930,6 +1937,7 @@ describe('generateText', () => {
"ai.model.provider": "mock-provider",
"ai.operationId": "ai.generateText",
"ai.prompt": "{"prompt":"test-input"}",
+ "ai.request.headers.user-agent": "ai/0.0.0-test",
"ai.response.finishReason": "stop",
"ai.response.toolCalls": "[{"toolCallId":"call-1","toolName":"tool1","input":"{ \\"value\\": \\"value\\" }"}]",
"ai.settings.maxRetries": 2,
@@ -1950,6 +1958,7 @@ describe('generateText', () => {
"ai.prompt.tools": [
"{"type":"function","name":"tool1","inputSchema":{"$schema":"http://json-schema.org/draft-07/schema#","type":"object","properties":{"value":{"type":"string"}},"required":["value"],"additionalProperties":false}}",
],
+ "ai.request.headers.user-agent": "ai/0.0.0-test",
"ai.response.finishReason": "stop",
"ai.response.id": "test-id",
"ai.response.model": "mock-model-id",
@@ -2266,7 +2275,7 @@ describe('generateText', () => {
});
describe('provider-executed tools', () => {
- describe('single provider-executed tool call and result', () => {
+ describe('two provider-executed tool calls and results', () => {
let result: GenerateTextResult;
beforeEach(async () => {
@@ -2370,6 +2379,51 @@ describe('generateText', () => {
`);
});
+ it('should include provider-executed tool calls in staticToolCalls', async () => {
+ expect(result.staticToolCalls).toMatchInlineSnapshot(`
+ [
+ {
+ "input": {
+ "value": "value",
+ },
+ "providerExecuted": true,
+ "providerMetadata": undefined,
+ "toolCallId": "call-1",
+ "toolName": "web_search",
+ "type": "tool-call",
+ },
+ {
+ "input": {
+ "value": "value",
+ },
+ "providerExecuted": true,
+ "providerMetadata": undefined,
+ "toolCallId": "call-2",
+ "toolName": "web_search",
+ "type": "tool-call",
+ },
+ ]
+ `);
+ });
+
+ it('should include provider-executed results in staticToolResults (errors excluded)', async () => {
+ expect(result.staticToolResults).toMatchInlineSnapshot(`
+ [
+ {
+ "dynamic": undefined,
+ "input": {
+ "value": "value",
+ },
+ "output": "{ "value": "result1" }",
+ "providerExecuted": true,
+ "toolCallId": "call-1",
+ "toolName": "web_search",
+ "type": "tool-result",
+ },
+ ]
+ `);
+ });
+
it('should only execute a single step', async () => {
expect(result.steps.length).toBe(1);
});
@@ -2470,37 +2524,39 @@ describe('generateText', () => {
});
expect(callOptions!).toMatchInlineSnapshot(`
- {
- "abortSignal": undefined,
- "frequencyPenalty": undefined,
- "headers": undefined,
- "maxOutputTokens": undefined,
- "presencePenalty": undefined,
- "prompt": [
- {
- "content": [
- {
- "text": "prompt",
- "type": "text",
- },
- ],
- "providerOptions": undefined,
- "role": "user",
- },
- ],
- "providerOptions": undefined,
- "responseFormat": {
- "type": "text",
- },
- "seed": undefined,
- "stopSequences": undefined,
- "temperature": undefined,
- "toolChoice": undefined,
- "tools": undefined,
- "topK": undefined,
- "topP": undefined,
- }
- `);
+ {
+ "abortSignal": undefined,
+ "frequencyPenalty": undefined,
+ "headers": {
+ "user-agent": "ai/0.0.0-test",
+ },
+ "maxOutputTokens": undefined,
+ "presencePenalty": undefined,
+ "prompt": [
+ {
+ "content": [
+ {
+ "text": "prompt",
+ "type": "text",
+ },
+ ],
+ "providerOptions": undefined,
+ "role": "user",
+ },
+ ],
+ "providerOptions": undefined,
+ "responseFormat": {
+ "type": "text",
+ },
+ "seed": undefined,
+ "stopSequences": undefined,
+ "temperature": undefined,
+ "toolChoice": undefined,
+ "tools": undefined,
+ "topK": undefined,
+ "topP": undefined,
+ }
+ `);
});
});
@@ -2542,50 +2598,52 @@ describe('generateText', () => {
});
expect(callOptions!).toMatchInlineSnapshot(`
- {
- "abortSignal": undefined,
- "frequencyPenalty": undefined,
- "headers": undefined,
- "maxOutputTokens": undefined,
- "presencePenalty": undefined,
- "prompt": [
- {
- "content": [
- {
- "text": "prompt",
- "type": "text",
- },
- ],
- "providerOptions": undefined,
- "role": "user",
- },
- ],
- "providerOptions": undefined,
- "responseFormat": {
- "schema": {
- "$schema": "http://json-schema.org/draft-07/schema#",
- "additionalProperties": false,
- "properties": {
- "value": {
- "type": "string",
+ {
+ "abortSignal": undefined,
+ "frequencyPenalty": undefined,
+ "headers": {
+ "user-agent": "ai/0.0.0-test",
+ },
+ "maxOutputTokens": undefined,
+ "presencePenalty": undefined,
+ "prompt": [
+ {
+ "content": [
+ {
+ "text": "prompt",
+ "type": "text",
+ },
+ ],
+ "providerOptions": undefined,
+ "role": "user",
+ },
+ ],
+ "providerOptions": undefined,
+ "responseFormat": {
+ "schema": {
+ "$schema": "http://json-schema.org/draft-07/schema#",
+ "additionalProperties": false,
+ "properties": {
+ "value": {
+ "type": "string",
+ },
},
+ "required": [
+ "value",
+ ],
+ "type": "object",
},
- "required": [
- "value",
- ],
- "type": "object",
- },
- "type": "json",
- },
- "seed": undefined,
- "stopSequences": undefined,
- "temperature": undefined,
- "toolChoice": undefined,
- "tools": undefined,
- "topK": undefined,
- "topP": undefined,
- }
- `);
+ "type": "json",
+ },
+ "seed": undefined,
+ "stopSequences": undefined,
+ "temperature": undefined,
+ "toolChoice": undefined,
+ "tools": undefined,
+ "topK": undefined,
+ "topP": undefined,
+ }
+ `);
});
});
});
diff --git a/packages/ai/src/generate-text/generate-text.ts b/packages/ai/src/generate-text/generate-text.ts
index ebfbb3c944b5..94a87ce382ca 100644
--- a/packages/ai/src/generate-text/generate-text.ts
+++ b/packages/ai/src/generate-text/generate-text.ts
@@ -9,6 +9,7 @@ import {
getErrorMessage,
IdGenerator,
ProviderOptions,
+ withUserAgentSuffix,
} from '@ai-sdk/provider-utils';
import { Tracer } from '@opentelemetry/api';
import { NoOutputSpecifiedError } from '../error/no-output-specified-error';
@@ -55,6 +56,7 @@ import { TypedToolError } from './tool-error';
import { ToolOutput } from './tool-output';
import { TypedToolResult } from './tool-result';
import { ToolSet } from './tool-set';
+import { VERSION } from '../version';
const originalGenerateId = createIdGenerator({
prefix: 'aitxt',
@@ -255,10 +257,15 @@ A function that attempts to repair a tool call that failed to parse.
const callSettings = prepareCallSettings(settings);
+ const headersWithUserAgent = withUserAgentSuffix(
+ headers ?? {},
+ `ai/${VERSION}`,
+ );
+
const baseTelemetryAttributes = getBaseTelemetryAttributes({
model,
telemetry,
- headers,
+ headers: headersWithUserAgent,
settings: { ...callSettings, maxRetries },
});
@@ -387,7 +394,7 @@ A function that attempts to repair a tool call that failed to parse.
prompt: promptMessages,
providerOptions,
abortSignal,
- headers,
+ headers: headersWithUserAgent,
});
// Fill in default values:
diff --git a/packages/ai/src/generate-text/index.ts b/packages/ai/src/generate-text/index.ts
index 5f36dd698cd2..ba890f16d76d 100644
--- a/packages/ai/src/generate-text/index.ts
+++ b/packages/ai/src/generate-text/index.ts
@@ -7,6 +7,7 @@ export type {
} from './generated-file';
export * as Output from './output';
export type { PrepareStepFunction, PrepareStepResult } from './prepare-step';
+export type { ReasoningOutput } from './reasoning-output';
export { smoothStream, type ChunkDetector } from './smooth-stream';
export type { StepResult } from './step-result';
export { hasToolCall, stepCountIs, type StopCondition } from './stop-condition';
diff --git a/packages/ai/src/generate-text/reasoning-output.ts b/packages/ai/src/generate-text/reasoning-output.ts
new file mode 100644
index 000000000000..85c03103bc34
--- /dev/null
+++ b/packages/ai/src/generate-text/reasoning-output.ts
@@ -0,0 +1,20 @@
+import { ProviderMetadata } from '../types/provider-metadata';
+
+/**
+ * Reasoning output of a text generation. It contains a reasoning.
+ */
+export interface ReasoningOutput {
+ type: 'reasoning';
+
+ /**
+ * The reasoning text.
+ */
+ text: string;
+
+ /**
+ * Additional provider-specific metadata. They are passed through
+ * to the provider from the AI SDK and enable provider-specific
+ * functionality that can be fully encapsulated in the provider.
+ */
+ providerMetadata?: ProviderMetadata;
+}
diff --git a/packages/ai/src/generate-text/step-result.ts b/packages/ai/src/generate-text/step-result.ts
index 6ea9f4c688f4..5c5ec1004263 100644
--- a/packages/ai/src/generate-text/step-result.ts
+++ b/packages/ai/src/generate-text/step-result.ts
@@ -199,7 +199,7 @@ export class DefaultStepResult
get staticToolCalls() {
return this.toolCalls.filter(
(toolCall): toolCall is StaticToolCall =>
- toolCall.dynamic === false,
+ toolCall.dynamic !== true,
);
}
@@ -216,7 +216,7 @@ export class DefaultStepResult
get staticToolResults() {
return this.toolResults.filter(
(toolResult): toolResult is StaticToolResult =>
- toolResult.dynamic === false,
+ toolResult.dynamic !== true,
);
}
diff --git a/packages/ai/src/generate-text/stream-text-result.ts b/packages/ai/src/generate-text/stream-text-result.ts
index 6080a9e42510..236d0e5cb422 100644
--- a/packages/ai/src/generate-text/stream-text-result.ts
+++ b/packages/ai/src/generate-text/stream-text-result.ts
@@ -1,10 +1,5 @@
-import { IdGenerator, ReasoningPart } from '@ai-sdk/provider-utils';
+import { IdGenerator } from '@ai-sdk/provider-utils';
import { ServerResponse } from 'node:http';
-import { InferUIMessageChunk } from '../ui-message-stream/ui-message-chunks';
-import { UIMessageStreamResponseInit } from '../ui-message-stream/ui-message-stream-response-init';
-import { InferUIMessageMetadata, UIMessage } from '../ui/ui-messages';
-import { AsyncIterableStream } from '../util/async-iterable-stream';
-import { ErrorHandler } from '../util/error-handler';
import {
CallWarning,
FinishReason,
@@ -14,9 +9,15 @@ import {
import { Source } from '../types/language-model';
import { LanguageModelResponseMetadata } from '../types/language-model-response-metadata';
import { LanguageModelUsage } from '../types/usage';
+import { InferUIMessageChunk } from '../ui-message-stream/ui-message-chunks';
import { UIMessageStreamOnFinishCallback } from '../ui-message-stream/ui-message-stream-on-finish-callback';
+import { UIMessageStreamResponseInit } from '../ui-message-stream/ui-message-stream-response-init';
+import { InferUIMessageMetadata, UIMessage } from '../ui/ui-messages';
+import { AsyncIterableStream } from '../util/async-iterable-stream';
+import { ErrorHandler } from '../util/error-handler';
import { ContentPart } from './content-part';
import { GeneratedFile } from './generated-file';
+import { ReasoningOutput } from './reasoning-output';
import { ResponseMessage } from './response-message';
import { StepResult } from './step-result';
import { DynamicToolCall, StaticToolCall, TypedToolCall } from './tool-call';
@@ -117,7 +118,7 @@ The full reasoning that the model has generated.
Automatically consumes the stream.
*/
- readonly reasoning: Promise>;
+ readonly reasoning: Promise>;
/**
The reasoning that has been generated by the last step.
diff --git a/packages/ai/src/generate-text/stream-text.test.ts b/packages/ai/src/generate-text/stream-text.test.ts
index 24b562f82efc..ccb3b3091209 100644
--- a/packages/ai/src/generate-text/stream-text.test.ts
+++ b/packages/ai/src/generate-text/stream-text.test.ts
@@ -4824,8 +4824,31 @@ describe('streamText', () => {
"timestamp": 1970-01-01T00:00:00.000Z,
},
"sources": [],
- "staticToolCalls": [],
- "staticToolResults": [],
+ "staticToolCalls": [
+ {
+ "input": {
+ "value": "value",
+ },
+ "providerExecuted": undefined,
+ "providerMetadata": undefined,
+ "toolCallId": "call-1",
+ "toolName": "tool1",
+ "type": "tool-call",
+ },
+ ],
+ "staticToolResults": [
+ {
+ "input": {
+ "value": "value",
+ },
+ "output": "value-result",
+ "providerExecuted": undefined,
+ "providerMetadata": undefined,
+ "toolCallId": "call-1",
+ "toolName": "tool1",
+ "type": "tool-result",
+ },
+ ],
"steps": [
DefaultStepResult {
"content": [
@@ -10623,8 +10646,31 @@ describe('streamText', () => {
"timestamp": 1970-01-01T00:00:00.000Z,
},
"sources": [],
- "staticToolCalls": [],
- "staticToolResults": [],
+ "staticToolCalls": [
+ {
+ "input": {
+ "value": "VALUE",
+ },
+ "providerExecuted": undefined,
+ "providerMetadata": undefined,
+ "toolCallId": "call-1",
+ "toolName": "tool1",
+ "type": "tool-call",
+ },
+ ],
+ "staticToolResults": [
+ {
+ "input": {
+ "value": "VALUE",
+ },
+ "output": "VALUE-RESULT",
+ "providerExecuted": undefined,
+ "providerMetadata": undefined,
+ "toolCallId": "call-1",
+ "toolName": "tool1",
+ "type": "tool-result",
+ },
+ ],
"steps": [
DefaultStepResult {
"content": [
diff --git a/packages/ai/src/global.ts b/packages/ai/src/global.ts
index 5bc6c6ffc1aa..72557a9a1686 100644
--- a/packages/ai/src/global.ts
+++ b/packages/ai/src/global.ts
@@ -1,4 +1,4 @@
-import { ProviderV2 } from '@ai-sdk/provider';
+import { ProviderV3 } from '@ai-sdk/provider';
import { LogWarningsFunction } from './logger/log-warnings';
// add AI SDK default provider to the globalThis object
@@ -11,7 +11,7 @@ declare global {
*
* @see https://ai-sdk.dev/docs/ai-sdk-core/provider-management#global-provider-configuration
*/
- var AI_SDK_DEFAULT_PROVIDER: ProviderV2 | undefined;
+ var AI_SDK_DEFAULT_PROVIDER: ProviderV3 | undefined;
/**
* The warning logger to use for the AI SDK.
diff --git a/packages/ai/src/index.ts b/packages/ai/src/index.ts
index 6d4cfca5a69f..d842c896f241 100644
--- a/packages/ai/src/index.ts
+++ b/packages/ai/src/index.ts
@@ -6,6 +6,7 @@ export {
dynamicTool,
generateId,
jsonSchema,
+ parseJsonEventStream,
tool,
zodSchema,
type IdGenerator,
diff --git a/packages/ai/src/logger/log-warnings.test.ts b/packages/ai/src/logger/log-warnings.test.ts
index 43873aee310c..5a748386dd36 100644
--- a/packages/ai/src/logger/log-warnings.test.ts
+++ b/packages/ai/src/logger/log-warnings.test.ts
@@ -7,7 +7,7 @@ import {
} from './log-warnings';
import type {
LanguageModelV2CallWarning,
- ImageModelV2CallWarning,
+ ImageModelV3CallWarning,
SpeechModelV2CallWarning,
TranscriptionModelV2CallWarning,
} from '@ai-sdk/provider';
@@ -55,7 +55,7 @@ describe('logWarnings', () => {
{
type: 'other',
message: 'Test warning 2',
- } as ImageModelV2CallWarning,
+ } as ImageModelV3CallWarning,
];
logWarnings(warnings);
@@ -96,7 +96,7 @@ describe('logWarnings', () => {
{
type: 'other',
message: 'Another warning',
- } as ImageModelV2CallWarning,
+ } as ImageModelV3CallWarning,
];
logWarnings(warnings);
@@ -141,7 +141,7 @@ describe('logWarnings', () => {
type: 'other',
message: 'First warning',
};
- const warning2: ImageModelV2CallWarning = {
+ const warning2: ImageModelV3CallWarning = {
type: 'unsupported-setting',
setting: 'size',
details: 'Size parameter not supported',
@@ -210,8 +210,8 @@ describe('logWarnings', () => {
);
});
- it('should log ImageModelV2CallWarning', () => {
- const warning: ImageModelV2CallWarning = {
+ it('should log image model call warning', () => {
+ const warning: ImageModelV3CallWarning = {
type: 'unsupported-setting',
setting: 'size',
details: 'Image size setting not supported',
@@ -266,7 +266,7 @@ describe('logWarnings', () => {
type: 'other',
message: 'Language model warning',
};
- const imageWarning: ImageModelV2CallWarning = {
+ const imageWarning: ImageModelV3CallWarning = {
type: 'other',
message: 'Image model warning',
};
diff --git a/packages/ai/src/logger/log-warnings.ts b/packages/ai/src/logger/log-warnings.ts
index a283cb143700..fe460aa8c70b 100644
--- a/packages/ai/src/logger/log-warnings.ts
+++ b/packages/ai/src/logger/log-warnings.ts
@@ -1,5 +1,5 @@
import {
- ImageModelV2CallWarning,
+ ImageModelV3CallWarning,
LanguageModelV2CallWarning,
SpeechModelV2CallWarning,
TranscriptionModelV2CallWarning,
@@ -7,7 +7,7 @@ import {
export type Warning =
| LanguageModelV2CallWarning
- | ImageModelV2CallWarning
+ | ImageModelV3CallWarning
| SpeechModelV2CallWarning
| TranscriptionModelV2CallWarning;
diff --git a/packages/ai/src/middleware/wrap-provider.test.ts b/packages/ai/src/middleware/wrap-provider.test.ts
index 7d6efb7faae5..6b2cda1769f8 100644
--- a/packages/ai/src/middleware/wrap-provider.test.ts
+++ b/packages/ai/src/middleware/wrap-provider.test.ts
@@ -1,5 +1,5 @@
import { MockLanguageModelV2 } from '../test/mock-language-model-v2';
-import { MockProviderV2 } from '../test/mock-provider-v2';
+import { MockProviderV3 } from '../test/mock-provider-v3';
import { wrapProvider } from './wrap-provider';
import { describe, it, expect, vi } from 'vitest';
@@ -9,7 +9,7 @@ describe('wrapProvider', () => {
const model2 = new MockLanguageModelV2({ modelId: 'model-2' });
const model3 = new MockLanguageModelV2({ modelId: 'model-3' });
- const provider = new MockProviderV2({
+ const provider = new MockProviderV3({
languageModels: {
'model-1': model1,
'model-2': model2,
diff --git a/packages/ai/src/middleware/wrap-provider.ts b/packages/ai/src/middleware/wrap-provider.ts
index b57151e66331..2af57aee5f34 100644
--- a/packages/ai/src/middleware/wrap-provider.ts
+++ b/packages/ai/src/middleware/wrap-provider.ts
@@ -1,25 +1,25 @@
-import type { ProviderV2 } from '@ai-sdk/provider';
+import type { ProviderV3 } from '@ai-sdk/provider';
import { LanguageModelMiddleware } from '../types/language-model-middleware';
import { wrapLanguageModel } from './wrap-language-model';
/**
- * Wraps a ProviderV2 instance with middleware functionality.
+ * Wraps a ProviderV3 instance with middleware functionality.
* This function allows you to apply middleware to all language models
* from the provider, enabling you to transform parameters, wrap generate
* operations, and wrap stream operations for every language model.
*
* @param options - Configuration options for wrapping the provider.
- * @param options.provider - The original ProviderV2 instance to be wrapped.
+ * @param options.provider - The original ProviderV3 instance to be wrapped.
* @param options.languageModelMiddleware - The middleware to be applied to all language models from the provider. When multiple middlewares are provided, the first middleware will transform the input first, and the last middleware will be wrapped directly around the model.
- * @returns A new ProviderV2 instance with middleware applied to all language models.
+ * @returns A new ProviderV3 instance with middleware applied to all language models.
*/
export function wrapProvider({
provider,
languageModelMiddleware,
}: {
- provider: ProviderV2;
+ provider: ProviderV3;
languageModelMiddleware: LanguageModelMiddleware | LanguageModelMiddleware[];
-}): ProviderV2 {
+}): ProviderV3 {
const wrappedProvider = {
languageModel(modelId: string) {
let model = provider.languageModel(modelId);
diff --git a/packages/ai/src/model/resolve-model.test.ts b/packages/ai/src/model/resolve-model.test.ts
index f47de854c78f..6f2c85539974 100644
--- a/packages/ai/src/model/resolve-model.test.ts
+++ b/packages/ai/src/model/resolve-model.test.ts
@@ -1,5 +1,5 @@
import { customProvider } from '../registry/custom-provider';
-import { MockEmbeddingModelV2 } from '../test/mock-embedding-model-v2';
+import { MockEmbeddingModelV3 } from '../test/mock-embedding-model-v3';
import { MockLanguageModelV2 } from '../test/mock-language-model-v2';
import { resolveEmbeddingModel, resolveLanguageModel } from './resolve-model';
import { beforeEach, afterEach, describe, expect, it } from 'vitest';
@@ -57,7 +57,7 @@ describe('resolveEmbeddingModel', () => {
describe('when a embedding model v2 is provided', () => {
it('should return the embedding model v2', () => {
const resolvedModel = resolveEmbeddingModel(
- new MockEmbeddingModelV2({
+ new MockEmbeddingModelV3({
provider: 'test-provider',
modelId: 'test-model-id',
}),
@@ -81,7 +81,7 @@ describe('resolveEmbeddingModel', () => {
beforeEach(() => {
globalThis.AI_SDK_DEFAULT_PROVIDER = customProvider({
textEmbeddingModels: {
- 'test-model-id': new MockEmbeddingModelV2({
+ 'test-model-id': new MockEmbeddingModelV3({
provider: 'global-test-provider',
modelId: 'actual-test-model-id',
}),
diff --git a/packages/ai/src/model/resolve-model.ts b/packages/ai/src/model/resolve-model.ts
index cd95bf217041..4cdb0fc20f82 100644
--- a/packages/ai/src/model/resolve-model.ts
+++ b/packages/ai/src/model/resolve-model.ts
@@ -1,8 +1,8 @@
import { gateway } from '@ai-sdk/gateway';
import {
- EmbeddingModelV2,
+ EmbeddingModelV3,
LanguageModelV2,
- ProviderV2,
+ ProviderV3,
} from '@ai-sdk/provider';
import { UnsupportedModelVersionError } from '../error';
import { EmbeddingModel } from '../types/embedding-model';
@@ -26,9 +26,9 @@ export function resolveLanguageModel(model: LanguageModel): LanguageModelV2 {
export function resolveEmbeddingModel(
model: EmbeddingModel,
-): EmbeddingModelV2 {
+): EmbeddingModelV3 {
if (typeof model !== 'string') {
- if (model.specificationVersion !== 'v2') {
+ if (model.specificationVersion !== 'v3') {
throw new UnsupportedModelVersionError({
version: model.specificationVersion,
provider: model.provider,
@@ -42,9 +42,9 @@ export function resolveEmbeddingModel(
// TODO AI SDK 6: figure out how to cleanly support different generic types
return getGlobalProvider().textEmbeddingModel(
model,
- ) as EmbeddingModelV2;
+ ) as EmbeddingModelV3;
}
-function getGlobalProvider(): ProviderV2 {
+function getGlobalProvider(): ProviderV3 {
return globalThis.AI_SDK_DEFAULT_PROVIDER ?? gateway;
}
diff --git a/packages/ai/src/prompt/convert-to-language-model-prompt.test.ts b/packages/ai/src/prompt/convert-to-language-model-prompt.test.ts
index 80b1ce7608ff..5401b1b65366 100644
--- a/packages/ai/src/prompt/convert-to-language-model-prompt.test.ts
+++ b/packages/ai/src/prompt/convert-to-language-model-prompt.test.ts
@@ -673,6 +673,101 @@ describe('convertToLanguageModelPrompt', () => {
]);
});
});
+
+ it('should download files when intermediate file cannot be downloaded', async () => {
+ const imageUrlA = `http://example.com/my-image-A.png`; // supported
+ const fileUrl = `http://127.0.0.1:3000/file`; // unsupported
+ const imageUrlB = `http://example.com/my-image-B.png`; // supported
+
+ const mockDownload = vi.fn().mockResolvedValue([
+ {
+ url: new URL(imageUrlA),
+ data: new Uint8Array([137, 80, 78, 71, 13, 10, 26, 10, 0]), // empty png and 0
+ mediaType: 'image/png',
+ },
+ null,
+ {
+ url: new URL(imageUrlB),
+ data: new Uint8Array([137, 80, 78, 71, 13, 10, 26, 10, 1]), // empty png and 1
+ mediaType: 'image/png',
+ },
+ ]);
+
+ const result = await convertToLanguageModelPrompt({
+ prompt: {
+ messages: [
+ {
+ role: 'user',
+ content: [
+ { type: 'image', image: imageUrlA, mediaType: 'image/png' },
+ {
+ type: 'file',
+ data: new URL(fileUrl),
+ mediaType: 'application/octet-stream',
+ },
+ { type: 'image', image: imageUrlB, mediaType: 'image/png' },
+ ],
+ },
+ ],
+ },
+ supportedUrls: {
+ '*': [/^https:\/\/.*$/],
+ },
+ download: mockDownload,
+ });
+
+ expect(result).toMatchInlineSnapshot(`
+ [
+ {
+ "content": [
+ {
+ "data": Uint8Array [
+ 137,
+ 80,
+ 78,
+ 71,
+ 13,
+ 10,
+ 26,
+ 10,
+ 0,
+ ],
+ "filename": undefined,
+ "mediaType": "image/png",
+ "providerOptions": undefined,
+ "type": "file",
+ },
+ {
+ "data": "http://127.0.0.1:3000/file",
+ "filename": undefined,
+ "mediaType": "application/octet-stream",
+ "providerOptions": undefined,
+ "type": "file",
+ },
+ {
+ "data": Uint8Array [
+ 137,
+ 80,
+ 78,
+ 71,
+ 13,
+ 10,
+ 26,
+ 10,
+ 1,
+ ],
+ "filename": undefined,
+ "mediaType": "image/png",
+ "providerOptions": undefined,
+ "type": "file",
+ },
+ ],
+ "providerOptions": undefined,
+ "role": "user",
+ },
+ ]
+ `);
+ });
});
describe('custom download function', () => {
@@ -899,7 +994,7 @@ describe('convertToLanguageModelMessage', () => {
describe('assistant message', () => {
describe('text parts', () => {
- it('should ignore empty text parts', async () => {
+ it('should ignore empty text parts when there are no provider options', async () => {
const result = convertToLanguageModelMessage({
message: {
role: 'assistant',
@@ -931,6 +1026,58 @@ describe('convertToLanguageModelMessage', () => {
],
});
});
+
+ it('should include empty text parts when there are provider options', async () => {
+ const result = convertToLanguageModelMessage({
+ message: {
+ role: 'assistant',
+ content: [
+ {
+ type: 'text',
+ text: '',
+ providerOptions: {
+ 'test-provider': {
+ 'key-a': 'test-value-1',
+ },
+ },
+ },
+ {
+ type: 'tool-call',
+ toolName: 'toolName',
+ toolCallId: 'toolCallId',
+ input: {},
+ },
+ ],
+ },
+ downloadedAssets: {},
+ });
+
+ expect(result).toMatchInlineSnapshot(`
+ {
+ "content": [
+ {
+ "providerOptions": {
+ "test-provider": {
+ "key-a": "test-value-1",
+ },
+ },
+ "text": "",
+ "type": "text",
+ },
+ {
+ "input": {},
+ "providerExecuted": undefined,
+ "providerOptions": undefined,
+ "toolCallId": "toolCallId",
+ "toolName": "toolName",
+ "type": "tool-call",
+ },
+ ],
+ "providerOptions": undefined,
+ "role": "assistant",
+ }
+ `);
+ });
});
describe('reasoning parts', () => {
diff --git a/packages/ai/src/prompt/convert-to-language-model-prompt.ts b/packages/ai/src/prompt/convert-to-language-model-prompt.ts
index f08152a4c3d4..13e726f4289e 100644
--- a/packages/ai/src/prompt/convert-to-language-model-prompt.ts
+++ b/packages/ai/src/prompt/convert-to-language-model-prompt.ts
@@ -108,8 +108,11 @@ export function convertToLanguageModelMessage({
role: 'assistant',
content: message.content
.filter(
- // remove empty text parts:
- part => part.type !== 'text' || part.text !== '',
+ // remove empty text parts (no text, and no provider options):
+ part =>
+ part.type !== 'text' ||
+ part.text !== '' ||
+ part.providerOptions != null,
)
.map(part => {
const providerOptions = part.providerOptions;
@@ -242,18 +245,15 @@ async function downloadAssets(
return Object.fromEntries(
downloadedFiles
- .filter(
- (
- downloadedFile,
- ): downloadedFile is {
- mediaType: string | undefined;
- data: Uint8Array;
- } => downloadedFile?.data != null,
+ .map((file, index) =>
+ file == null
+ ? null
+ : [
+ plannedDownloads[index].url.toString(),
+ { data: file.data, mediaType: file.mediaType },
+ ],
)
- .map(({ data, mediaType }, index) => [
- plannedDownloads[index].url.toString(),
- { data, mediaType },
- ]),
+ .filter(file => file != null),
);
}
diff --git a/packages/ai/src/registry/custom-provider.test.ts b/packages/ai/src/registry/custom-provider.test.ts
index 8a4a0986454a..e3b417161e58 100644
--- a/packages/ai/src/registry/custom-provider.test.ts
+++ b/packages/ai/src/registry/custom-provider.test.ts
@@ -1,14 +1,14 @@
import { NoSuchModelError } from '@ai-sdk/provider';
import { describe, expect, it, vi } from 'vitest';
-import { MockEmbeddingModelV2 } from '../test/mock-embedding-model-v2';
-import { MockImageModelV2 } from '../test/mock-image-model-v2';
+import { MockEmbeddingModelV3 } from '../test/mock-embedding-model-v3';
+import { MockImageModelV3 } from '../test/mock-image-model-v3';
import { MockLanguageModelV2 } from '../test/mock-language-model-v2';
import { MockTranscriptionModelV2 } from '../test/mock-transcription-model-v2';
import { MockSpeechModelV2 } from '../test/mock-speech-model-v2';
import { customProvider } from './custom-provider';
const mockLanguageModel = new MockLanguageModelV2();
-const mockEmbeddingModel = new MockEmbeddingModelV2();
+const mockEmbeddingModel = new MockEmbeddingModelV3();
const mockFallbackProvider = {
languageModel: vi.fn(),
textEmbeddingModel: vi.fn(),
@@ -79,7 +79,7 @@ describe('textEmbeddingModel', () => {
});
describe('imageModel', () => {
- const mockImageModel = new MockImageModelV2();
+ const mockImageModel = new MockImageModelV3();
it('should return the image model if it exists', () => {
const provider = customProvider({
diff --git a/packages/ai/src/registry/custom-provider.ts b/packages/ai/src/registry/custom-provider.ts
index b235d5e7c26c..8bc5fad834a6 100644
--- a/packages/ai/src/registry/custom-provider.ts
+++ b/packages/ai/src/registry/custom-provider.ts
@@ -1,9 +1,9 @@
import {
- EmbeddingModelV2,
- ImageModelV2,
+ EmbeddingModelV3,
+ ImageModelV3,
LanguageModelV2,
NoSuchModelError,
- ProviderV2,
+ ProviderV3,
SpeechModelV2,
TranscriptionModelV2,
} from '@ai-sdk/provider';
@@ -24,8 +24,8 @@ import {
*/
export function customProvider<
LANGUAGE_MODELS extends Record,
- EMBEDDING_MODELS extends Record>,
- IMAGE_MODELS extends Record,
+ EMBEDDING_MODELS extends Record>,
+ IMAGE_MODELS extends Record,
TRANSCRIPTION_MODELS extends Record,
SPEECH_MODELS extends Record,
>({
@@ -41,13 +41,13 @@ export function customProvider<
imageModels?: IMAGE_MODELS;
transcriptionModels?: TRANSCRIPTION_MODELS;
speechModels?: SPEECH_MODELS;
- fallbackProvider?: ProviderV2;
-}): ProviderV2 & {
+ fallbackProvider?: ProviderV3;
+}): ProviderV3 & {
languageModel(modelId: ExtractModelId): LanguageModelV2;
textEmbeddingModel(
modelId: ExtractModelId,
- ): EmbeddingModelV2;
- imageModel(modelId: ExtractModelId): ImageModelV2;
+ ): EmbeddingModelV3;
+ imageModel(modelId: ExtractModelId): ImageModelV3;
transcriptionModel(
modelId: ExtractModelId,
): TranscriptionModelV2;
@@ -68,7 +68,7 @@ export function customProvider<
textEmbeddingModel(
modelId: ExtractModelId,
- ): EmbeddingModelV2 {
+ ): EmbeddingModelV3 {
if (textEmbeddingModels != null && modelId in textEmbeddingModels) {
return textEmbeddingModels[modelId];
}
@@ -80,7 +80,7 @@ export function customProvider<
throw new NoSuchModelError({ modelId, modelType: 'textEmbeddingModel' });
},
- imageModel(modelId: ExtractModelId): ImageModelV2 {
+ imageModel(modelId: ExtractModelId): ImageModelV3 {
if (imageModels != null && modelId in imageModels) {
return imageModels[modelId];
}
diff --git a/packages/ai/src/registry/provider-registry.test.ts b/packages/ai/src/registry/provider-registry.test.ts
index 6ada32d2a854..6b23d5f84a7c 100644
--- a/packages/ai/src/registry/provider-registry.test.ts
+++ b/packages/ai/src/registry/provider-registry.test.ts
@@ -1,12 +1,12 @@
import { NoSuchModelError } from '@ai-sdk/provider';
-import { MockEmbeddingModelV2 } from '../test/mock-embedding-model-v2';
+import { MockEmbeddingModelV3 } from '../test/mock-embedding-model-v3';
import { MockLanguageModelV2 } from '../test/mock-language-model-v2';
import { NoSuchProviderError } from './no-such-provider-error';
import { createProviderRegistry } from './provider-registry';
-import { MockImageModelV2 } from '../test/mock-image-model-v2';
+import { MockImageModelV3 } from '../test/mock-image-model-v3';
import { MockTranscriptionModelV2 } from '../test/mock-transcription-model-v2';
import { MockSpeechModelV2 } from '../test/mock-speech-model-v2';
-import { MockProviderV2 } from '../test/mock-provider-v2';
+import { MockProviderV3 } from '../test/mock-provider-v3';
import { describe, it, expect, vi } from 'vitest';
describe('languageModel', () => {
@@ -159,7 +159,7 @@ describe('languageModel', () => {
describe('textEmbeddingModel', () => {
it('should return embedding model from provider using textEmbeddingModel', () => {
- const model = new MockEmbeddingModelV2();
+ const model = new MockEmbeddingModelV3();
const modelRegistry = createProviderRegistry({
provider: {
@@ -224,7 +224,7 @@ describe('textEmbeddingModel', () => {
});
it('should support custom separator', () => {
- const model = new MockEmbeddingModelV2();
+ const model = new MockEmbeddingModelV3();
const modelRegistry = createProviderRegistry(
{
@@ -256,7 +256,7 @@ describe('textEmbeddingModel', () => {
describe('imageModel', () => {
it('should return image model from provider', () => {
- const model = new MockImageModelV2();
+ const model = new MockImageModelV3();
const modelRegistry = createProviderRegistry({
provider: {
@@ -305,7 +305,7 @@ describe('imageModel', () => {
});
it('should support custom separator', () => {
- const model = new MockImageModelV2();
+ const model = new MockImageModelV3();
const modelRegistry = createProviderRegistry(
{
@@ -435,14 +435,14 @@ describe('middleware functionality', () => {
const model2 = new MockLanguageModelV2({ modelId: 'model-2' });
const model3 = new MockLanguageModelV2({ modelId: 'model-3' });
- const provider1 = new MockProviderV2({
+ const provider1 = new MockProviderV3({
languageModels: {
'model-1': model1,
'model-2': model2,
},
});
- const provider2 = new MockProviderV2({
+ const provider2 = new MockProviderV3({
languageModels: {
'model-3': model3,
},
diff --git a/packages/ai/src/registry/provider-registry.ts b/packages/ai/src/registry/provider-registry.ts
index 929b0d428c17..0e4f21f0425e 100644
--- a/packages/ai/src/registry/provider-registry.ts
+++ b/packages/ai/src/registry/provider-registry.ts
@@ -1,9 +1,9 @@
import {
- EmbeddingModelV2,
- ImageModelV2,
+ EmbeddingModelV3,
+ ImageModelV3,
LanguageModelV2,
NoSuchModelError,
- ProviderV2,
+ ProviderV3,
SpeechModelV2,
TranscriptionModelV2,
} from '@ai-sdk/provider';
@@ -18,7 +18,7 @@ type ExtractLiteralUnion = T extends string
: never;
export interface ProviderRegistryProvider<
- PROVIDERS extends Record = Record,
+ PROVIDERS extends Record = Record,
SEPARATOR extends string = ':',
> {
languageModel(
@@ -34,19 +34,19 @@ export interface ProviderRegistryProvider<
id: KEY extends string
? `${KEY & string}${SEPARATOR}${ExtractLiteralUnion>[0]>}`
: never,
- ): EmbeddingModelV2;
+ ): EmbeddingModelV3;
textEmbeddingModel(
id: KEY extends string ? `${KEY & string}${SEPARATOR}${string}` : never,
- ): EmbeddingModelV2;
+ ): EmbeddingModelV3;
imageModel(
id: KEY extends string
? `${KEY & string}${SEPARATOR}${ExtractLiteralUnion>[0]>}`
: never,
- ): ImageModelV2;
+ ): ImageModelV3;
imageModel(
id: KEY extends string ? `${KEY & string}${SEPARATOR}${string}` : never,
- ): ImageModelV2;
+ ): ImageModelV3;
transcriptionModel(
id: KEY extends string
@@ -80,7 +80,7 @@ export interface ProviderRegistryProvider<
* @returns A new ProviderRegistryProvider instance that provides access to all registered providers with optional middleware applied to language models.
*/
export function createProviderRegistry<
- PROVIDERS extends Record,
+ PROVIDERS extends Record,
SEPARATOR extends string = ':',
>(
providers: PROVIDERS,
@@ -115,7 +115,7 @@ export function createProviderRegistry<
export const experimental_createProviderRegistry = createProviderRegistry;
class DefaultProviderRegistry<
- PROVIDERS extends Record,
+ PROVIDERS extends Record,
SEPARATOR extends string,
> implements ProviderRegistryProvider
{
@@ -156,7 +156,7 @@ class DefaultProviderRegistry<
| 'imageModel'
| 'transcriptionModel'
| 'speechModel',
- ): ProviderV2 {
+ ): ProviderV3 {
const provider = this.providers[id as keyof PROVIDERS];
if (provider == null) {
@@ -219,7 +219,7 @@ class DefaultProviderRegistry<
textEmbeddingModel(
id: `${KEY & string}${SEPARATOR}${string}`,
- ): EmbeddingModelV2 {
+ ): EmbeddingModelV3 {
const [providerId, modelId] = this.splitId(id, 'textEmbeddingModel');
const provider = this.getProvider(providerId, 'textEmbeddingModel');
@@ -237,7 +237,7 @@ class DefaultProviderRegistry<
imageModel(
id: `${KEY & string}${SEPARATOR}${string}`,
- ): ImageModelV2 {
+ ): ImageModelV3 {
const [providerId, modelId] = this.splitId(id, 'imageModel');
const provider = this.getProvider(providerId, 'imageModel');
diff --git a/packages/ai/src/test/mock-embedding-model-v2.ts b/packages/ai/src/test/mock-embedding-model-v2.ts
deleted file mode 100644
index 0dccd1734de7..000000000000
--- a/packages/ai/src/test/mock-embedding-model-v2.ts
+++ /dev/null
@@ -1,35 +0,0 @@
-import { EmbeddingModelV2 } from '@ai-sdk/provider';
-import { notImplemented } from './not-implemented';
-
-export class MockEmbeddingModelV2 implements EmbeddingModelV2 {
- readonly specificationVersion = 'v2';
-
- readonly provider: EmbeddingModelV2['provider'];
- readonly modelId: EmbeddingModelV2['modelId'];
- readonly maxEmbeddingsPerCall: EmbeddingModelV2['maxEmbeddingsPerCall'];
- readonly supportsParallelCalls: EmbeddingModelV2['supportsParallelCalls'];
-
- doEmbed: EmbeddingModelV2['doEmbed'];
-
- constructor({
- provider = 'mock-provider',
- modelId = 'mock-model-id',
- maxEmbeddingsPerCall = 1,
- supportsParallelCalls = false,
- doEmbed = notImplemented,
- }: {
- provider?: EmbeddingModelV2['provider'];
- modelId?: EmbeddingModelV2['modelId'];
- maxEmbeddingsPerCall?:
- | EmbeddingModelV2['maxEmbeddingsPerCall']
- | null;
- supportsParallelCalls?: EmbeddingModelV2['supportsParallelCalls'];
- doEmbed?: EmbeddingModelV2['doEmbed'];
- } = {}) {
- this.provider = provider;
- this.modelId = modelId;
- this.maxEmbeddingsPerCall = maxEmbeddingsPerCall ?? undefined;
- this.supportsParallelCalls = supportsParallelCalls;
- this.doEmbed = doEmbed;
- }
-}
diff --git a/packages/ai/src/test/mock-embedding-model-v3.ts b/packages/ai/src/test/mock-embedding-model-v3.ts
new file mode 100644
index 000000000000..409a329c91b9
--- /dev/null
+++ b/packages/ai/src/test/mock-embedding-model-v3.ts
@@ -0,0 +1,35 @@
+import { EmbeddingModelV3 } from '@ai-sdk/provider';
+import { notImplemented } from './not-implemented';
+
+export class MockEmbeddingModelV3 implements EmbeddingModelV3 {
+ readonly specificationVersion = 'v3';
+
+ readonly provider: EmbeddingModelV3['provider'];
+ readonly modelId: EmbeddingModelV3['modelId'];
+ readonly maxEmbeddingsPerCall: EmbeddingModelV3['maxEmbeddingsPerCall'];
+ readonly supportsParallelCalls: EmbeddingModelV3['supportsParallelCalls'];
+
+ doEmbed: EmbeddingModelV3['doEmbed'];
+
+ constructor({
+ provider = 'mock-provider',
+ modelId = 'mock-model-id',
+ maxEmbeddingsPerCall = 1,
+ supportsParallelCalls = false,
+ doEmbed = notImplemented,
+ }: {
+ provider?: EmbeddingModelV3['provider'];
+ modelId?: EmbeddingModelV3['modelId'];
+ maxEmbeddingsPerCall?:
+ | EmbeddingModelV3['maxEmbeddingsPerCall']
+ | null;
+ supportsParallelCalls?: EmbeddingModelV3['supportsParallelCalls'];
+ doEmbed?: EmbeddingModelV3['doEmbed'];
+ } = {}) {
+ this.provider = provider;
+ this.modelId = modelId;
+ this.maxEmbeddingsPerCall = maxEmbeddingsPerCall ?? undefined;
+ this.supportsParallelCalls = supportsParallelCalls;
+ this.doEmbed = doEmbed;
+ }
+}
diff --git a/packages/ai/src/test/mock-image-model-v2.ts b/packages/ai/src/test/mock-image-model-v2.ts
deleted file mode 100644
index 73876b4c44ea..000000000000
--- a/packages/ai/src/test/mock-image-model-v2.ts
+++ /dev/null
@@ -1,28 +0,0 @@
-import { ImageModelV2 } from '@ai-sdk/provider';
-import { notImplemented } from './not-implemented';
-
-export class MockImageModelV2 implements ImageModelV2 {
- readonly specificationVersion = 'v2';
- readonly provider: ImageModelV2['provider'];
- readonly modelId: ImageModelV2['modelId'];
- readonly maxImagesPerCall: ImageModelV2['maxImagesPerCall'];
-
- doGenerate: ImageModelV2['doGenerate'];
-
- constructor({
- provider = 'mock-provider',
- modelId = 'mock-model-id',
- maxImagesPerCall = 1,
- doGenerate = notImplemented,
- }: {
- provider?: ImageModelV2['provider'];
- modelId?: ImageModelV2['modelId'];
- maxImagesPerCall?: ImageModelV2['maxImagesPerCall'];
- doGenerate?: ImageModelV2['doGenerate'];
- } = {}) {
- this.provider = provider;
- this.modelId = modelId;
- this.maxImagesPerCall = maxImagesPerCall;
- this.doGenerate = doGenerate;
- }
-}
diff --git a/packages/ai/src/test/mock-image-model-v3.ts b/packages/ai/src/test/mock-image-model-v3.ts
new file mode 100644
index 000000000000..96d599abd438
--- /dev/null
+++ b/packages/ai/src/test/mock-image-model-v3.ts
@@ -0,0 +1,28 @@
+import { ImageModelV3 } from '@ai-sdk/provider';
+import { notImplemented } from './not-implemented';
+
+export class MockImageModelV3 implements ImageModelV3 {
+ readonly specificationVersion = 'v3';
+ readonly provider: ImageModelV3['provider'];
+ readonly modelId: ImageModelV3['modelId'];
+ readonly maxImagesPerCall: ImageModelV3['maxImagesPerCall'];
+
+ doGenerate: ImageModelV3['doGenerate'];
+
+ constructor({
+ provider = 'mock-provider',
+ modelId = 'mock-model-id',
+ maxImagesPerCall = 1,
+ doGenerate = notImplemented,
+ }: {
+ provider?: ImageModelV3['provider'];
+ modelId?: ImageModelV3['modelId'];
+ maxImagesPerCall?: ImageModelV3['maxImagesPerCall'];
+ doGenerate?: ImageModelV3['doGenerate'];
+ } = {}) {
+ this.provider = provider;
+ this.modelId = modelId;
+ this.maxImagesPerCall = maxImagesPerCall;
+ this.doGenerate = doGenerate;
+ }
+}
diff --git a/packages/ai/src/test/mock-provider-v2.ts b/packages/ai/src/test/mock-provider-v3.ts
similarity index 77%
rename from packages/ai/src/test/mock-provider-v2.ts
rename to packages/ai/src/test/mock-provider-v3.ts
index a07974b1ea13..249c85265060 100644
--- a/packages/ai/src/test/mock-provider-v2.ts
+++ b/packages/ai/src/test/mock-provider-v3.ts
@@ -1,19 +1,19 @@
import {
- EmbeddingModelV2,
- ImageModelV2,
+ EmbeddingModelV3,
+ ImageModelV3,
LanguageModelV2,
NoSuchModelError,
- ProviderV2,
+ ProviderV3,
SpeechModelV2,
TranscriptionModelV2,
} from '@ai-sdk/provider';
-export class MockProviderV2 implements ProviderV2 {
- languageModel: ProviderV2['languageModel'];
- textEmbeddingModel: ProviderV2['textEmbeddingModel'];
- imageModel: ProviderV2['imageModel'];
- transcriptionModel: ProviderV2['transcriptionModel'];
- speechModel: ProviderV2['speechModel'];
+export class MockProviderV3 implements ProviderV3 {
+ languageModel: ProviderV3['languageModel'];
+ textEmbeddingModel: ProviderV3['textEmbeddingModel'];
+ imageModel: ProviderV3['imageModel'];
+ transcriptionModel: ProviderV3['transcriptionModel'];
+ speechModel: ProviderV3['speechModel'];
constructor({
languageModels,
@@ -23,8 +23,8 @@ export class MockProviderV2 implements ProviderV2 {
speechModels,
}: {
languageModels?: Record;
- embeddingModels?: Record>;
- imageModels?: Record;
+ embeddingModels?: Record>;
+ imageModels?: Record;
transcriptionModels?: Record;
speechModels?: Record;
} = {}) {
diff --git a/packages/ai/src/tool/mcp/mcp-sse-transport.test.ts b/packages/ai/src/tool/mcp/mcp-sse-transport.test.ts
index c6470b5eb59d..4fafb87718e4 100644
--- a/packages/ai/src/tool/mcp/mcp-sse-transport.test.ts
+++ b/packages/ai/src/tool/mcp/mcp-sse-transport.test.ts
@@ -1,7 +1,7 @@
import {
createTestServer,
TestResponseController,
-} from '@ai-sdk/provider-utils/test';
+} from '@ai-sdk/test-server/with-vitest';
import { MCPClientError } from '../../error/mcp-client-error';
import { SseMCPTransport } from './mcp-sse-transport';
import { beforeEach, describe, expect, it } from 'vitest';
@@ -269,12 +269,14 @@ describe('SseMCPTransport', () => {
accept: 'text/event-stream',
...customHeaders,
});
+ expect(server.calls[0].requestUserAgent).toContain('ai-sdk/');
// Verify POST request headers
expect(server.calls[1].requestHeaders).toEqual({
'content-type': 'application/json',
...customHeaders,
});
+ expect(server.calls[1].requestUserAgent).toContain('ai-sdk/');
await transport.close();
});
diff --git a/packages/ai/src/tool/mcp/mcp-sse-transport.ts b/packages/ai/src/tool/mcp/mcp-sse-transport.ts
index 1fd2ddc7ff14..5ca2be0b93f4 100644
--- a/packages/ai/src/tool/mcp/mcp-sse-transport.ts
+++ b/packages/ai/src/tool/mcp/mcp-sse-transport.ts
@@ -1,7 +1,12 @@
-import { EventSourceParserStream } from '@ai-sdk/provider-utils';
+import {
+ EventSourceParserStream,
+ withUserAgentSuffix,
+ getRuntimeEnvironmentUserAgent,
+} from '@ai-sdk/provider-utils';
import { MCPClientError } from '../../error/mcp-client-error';
import { JSONRPCMessage, JSONRPCMessageSchema } from './json-rpc-message';
import { MCPTransport } from './mcp-transport';
+import { VERSION } from '../../version';
export class SseMCPTransport implements MCPTransport {
private endpoint?: URL;
@@ -38,8 +43,14 @@ export class SseMCPTransport implements MCPTransport {
const establishConnection = async () => {
try {
- const headers = new Headers(this.headers);
- headers.set('Accept', 'text/event-stream');
+ const headers = withUserAgentSuffix(
+ {
+ ...this.headers,
+ Accept: 'text/event-stream',
+ },
+ `ai-sdk/${VERSION}`,
+ getRuntimeEnvironmentUserAgent(),
+ );
const response = await fetch(this.url.href, {
headers,
signal: this.abortController?.signal,
@@ -149,8 +160,14 @@ export class SseMCPTransport implements MCPTransport {
}
try {
- const headers = new Headers(this.headers);
- headers.set('Content-Type', 'application/json');
+ const headers = withUserAgentSuffix(
+ {
+ ...this.headers,
+ 'Content-Type': 'application/json',
+ },
+ `ai-sdk/${VERSION}`,
+ getRuntimeEnvironmentUserAgent(),
+ );
const init = {
method: 'POST',
headers,
diff --git a/packages/ai/src/transcribe/transcribe.test.ts b/packages/ai/src/transcribe/transcribe.test.ts
index 9be4f6ee6db1..9160da6bb7f0 100644
--- a/packages/ai/src/transcribe/transcribe.test.ts
+++ b/packages/ai/src/transcribe/transcribe.test.ts
@@ -3,11 +3,25 @@ import {
TranscriptionModelV2,
TranscriptionModelV2CallWarning,
} from '@ai-sdk/provider';
-import { afterEach, beforeEach, describe, expect, it, vitest } from 'vitest';
+import {
+ afterEach,
+ beforeEach,
+ describe,
+ expect,
+ it,
+ vitest,
+ vi,
+} from 'vitest';
import * as logWarningsModule from '../logger/log-warnings';
import { MockTranscriptionModelV2 } from '../test/mock-transcription-model-v2';
import { transcribe } from './transcribe';
+vi.mock('../version', () => {
+ return {
+ VERSION: '0.0.0-test',
+ };
+});
+
const audioData = new Uint8Array([1, 2, 3, 4]); // Sample audio data
const testDate = new Date(2024, 0, 1);
@@ -86,14 +100,19 @@ describe('transcribe', () => {
},
}),
audio: audioData,
- headers: { 'custom-request-header': 'request-header-value' },
+ headers: {
+ 'custom-request-header': 'request-header-value',
+ },
abortSignal,
});
expect(capturedArgs).toStrictEqual({
audio: audioData,
mediaType: 'audio/wav',
- headers: { 'custom-request-header': 'request-header-value' },
+ headers: {
+ 'custom-request-header': 'request-header-value',
+ 'user-agent': 'ai/0.0.0-test',
+ },
abortSignal,
providerOptions: {},
});
@@ -239,6 +258,7 @@ describe('transcribe', () => {
timestamp: testDate,
headers: {
'custom-response-header': 'response-header-value',
+ 'user-agent': 'ai/0.0.0-test',
},
}),
}),
@@ -253,6 +273,7 @@ describe('transcribe', () => {
modelId: expect.any(String),
headers: {
'custom-response-header': 'response-header-value',
+ 'user-agent': 'ai/0.0.0-test',
},
},
],
diff --git a/packages/ai/src/transcribe/transcribe.ts b/packages/ai/src/transcribe/transcribe.ts
index 7e36d8e4e13c..bbf19b89eb4d 100644
--- a/packages/ai/src/transcribe/transcribe.ts
+++ b/packages/ai/src/transcribe/transcribe.ts
@@ -1,5 +1,5 @@
import { JSONValue, TranscriptionModelV2 } from '@ai-sdk/provider';
-import { ProviderOptions } from '@ai-sdk/provider-utils';
+import { ProviderOptions, withUserAgentSuffix } from '@ai-sdk/provider-utils';
import { NoTranscriptGeneratedError } from '../error/no-transcript-generated-error';
import { UnsupportedModelVersionError } from '../error/unsupported-model-version-error';
import { logWarnings } from '../logger/log-warnings';
@@ -14,7 +14,7 @@ import {
import { download } from '../util/download/download';
import { prepareRetries } from '../util/prepare-retries';
import { TranscriptionResult } from './transcribe-result';
-
+import { VERSION } from '../version';
/**
Generates transcripts using a transcription model.
@@ -93,6 +93,11 @@ Only applicable for HTTP-based providers.
abortSignal,
});
+ const headersWithUserAgent = withUserAgentSuffix(
+ headers ?? {},
+ `ai/${VERSION}`,
+ );
+
const audioData =
audio instanceof URL
? (await download({ url: audio })).data
@@ -102,7 +107,7 @@ Only applicable for HTTP-based providers.
model.doGenerate({
audio: audioData,
abortSignal,
- headers,
+ headers: headersWithUserAgent,
providerOptions,
mediaType:
detectMediaType({
diff --git a/packages/ai/src/types/embedding-model.ts b/packages/ai/src/types/embedding-model.ts
index d2788c2ad5cc..6dd7d26a0cc8 100644
--- a/packages/ai/src/types/embedding-model.ts
+++ b/packages/ai/src/types/embedding-model.ts
@@ -1,11 +1,11 @@
-import { EmbeddingModelV2, EmbeddingModelV2Embedding } from '@ai-sdk/provider';
+import { EmbeddingModelV3, EmbeddingModelV3Embedding } from '@ai-sdk/provider';
/**
Embedding model that is used by the AI SDK Core functions.
*/
-export type EmbeddingModel = string | EmbeddingModelV2;
+export type EmbeddingModel = string | EmbeddingModelV3;
/**
Embedding.
*/
-export type Embedding = EmbeddingModelV2Embedding;
+export type Embedding = EmbeddingModelV3Embedding;
diff --git a/packages/ai/src/types/image-model.ts b/packages/ai/src/types/image-model.ts
index efec355395f8..e8ab40cfc376 100644
--- a/packages/ai/src/types/image-model.ts
+++ b/packages/ai/src/types/image-model.ts
@@ -1,21 +1,21 @@
import {
- ImageModelV2,
- ImageModelV2CallWarning,
- ImageModelV2ProviderMetadata,
+ ImageModelV3,
+ ImageModelV3CallWarning,
+ ImageModelV3ProviderMetadata,
} from '@ai-sdk/provider';
/**
Image model that is used by the AI SDK Core functions.
*/
-export type ImageModel = ImageModelV2;
+export type ImageModel = ImageModelV3;
/**
Warning from the model provider for this call. The call will proceed, but e.g.
some settings might not be supported, which can lead to suboptimal results.
*/
-export type ImageGenerationWarning = ImageModelV2CallWarning;
+export type ImageGenerationWarning = ImageModelV3CallWarning;
/**
Metadata from the model provider for this call
*/
-export type ImageModelProviderMetadata = ImageModelV2ProviderMetadata;
+export type ImageModelProviderMetadata = ImageModelV3ProviderMetadata;
diff --git a/packages/ai/src/types/usage.ts b/packages/ai/src/types/usage.ts
index af045f39b014..4f82d6b274f4 100644
--- a/packages/ai/src/types/usage.ts
+++ b/packages/ai/src/types/usage.ts
@@ -8,7 +8,7 @@ export type LanguageModelUsage = LanguageModelV2Usage;
/**
Represents the number of tokens used in an embedding.
*/
-// TODO replace with EmbeddingModelV2Usage
+// TODO replace with EmbeddingModelV3Usage
export type EmbeddingModelUsage = {
/**
The number of tokens used in the embedding.
diff --git a/packages/ai/src/ui-message-stream/index.ts b/packages/ai/src/ui-message-stream/index.ts
index 29da2f00ff7b..83876a7ac744 100644
--- a/packages/ai/src/ui-message-stream/index.ts
+++ b/packages/ai/src/ui-message-stream/index.ts
@@ -3,7 +3,11 @@ export { createUIMessageStreamResponse } from './create-ui-message-stream-respon
export { JsonToSseTransformStream } from './json-to-sse-transform-stream';
export { pipeUIMessageStreamToResponse } from './pipe-ui-message-stream-to-response';
export { readUIMessageStream } from './read-ui-message-stream';
-export type { InferUIMessageChunk, UIMessageChunk } from './ui-message-chunks';
+export {
+ uiMessageChunkSchema,
+ type InferUIMessageChunk,
+ type UIMessageChunk,
+} from './ui-message-chunks';
export { UI_MESSAGE_STREAM_HEADERS } from './ui-message-stream-headers';
export type { UIMessageStreamOnFinishCallback } from './ui-message-stream-on-finish-callback';
export type { UIMessageStreamWriter } from './ui-message-stream-writer';
diff --git a/packages/ai/src/ui-message-stream/ui-message-chunks.test-d.ts b/packages/ai/src/ui-message-stream/ui-message-chunks.test-d.ts
new file mode 100644
index 000000000000..df83b5da3039
--- /dev/null
+++ b/packages/ai/src/ui-message-stream/ui-message-chunks.test-d.ts
@@ -0,0 +1,14 @@
+import { describe, expectTypeOf, it } from 'vitest';
+import { UIMessageChunk, uiMessageChunkSchema } from './ui-message-chunks';
+
+describe('UI message chunks type', () => {
+ it('parsed UI message chunk should have UIMessageChunk type', () => {
+ const chunk = uiMessageChunkSchema.parse({
+ type: 'text-delta',
+ delta: 'Hello, world!',
+ id: '123',
+ });
+
+ expectTypeOf(chunk).toEqualTypeOf();
+ });
+});
diff --git a/packages/ai/src/ui-message-stream/ui-message-chunks.ts b/packages/ai/src/ui-message-stream/ui-message-chunks.ts
index ebda56cc21d5..aad82429aa62 100644
--- a/packages/ai/src/ui-message-stream/ui-message-chunks.ts
+++ b/packages/ai/src/ui-message-stream/ui-message-chunks.ts
@@ -78,11 +78,6 @@ export const uiMessageChunkSchema = z.union([
providerExecuted: z.boolean().optional(),
dynamic: z.boolean().optional(),
}),
- z.strictObject({
- type: z.literal('reasoning'),
- text: z.string(),
- providerMetadata: providerMetadataSchema.optional(),
- }),
z.strictObject({
type: z.literal('reasoning-start'),
id: z.string(),
@@ -99,9 +94,6 @@ export const uiMessageChunkSchema = z.union([
id: z.string(),
providerMetadata: providerMetadataSchema.optional(),
}),
- z.strictObject({
- type: z.literal('reasoning-part-finish'),
- }),
z.strictObject({
type: z.literal('source-url'),
sourceId: z.string(),
@@ -124,7 +116,11 @@ export const uiMessageChunkSchema = z.union([
providerMetadata: providerMetadataSchema.optional(),
}),
z.strictObject({
- type: z.string().startsWith('data-'),
+ type: z.custom<`data-${string}`>(
+ (value): value is `data-${string}` =>
+ typeof value === 'string' && value.startsWith('data-'),
+ { message: 'Type must start with "data-"' },
+ ),
id: z.string().optional(),
data: z.unknown(),
transient: z.boolean().optional(),
@@ -267,6 +263,7 @@ export type UIMessageChunk<
type: 'file';
url: string;
mediaType: string;
+ providerMetadata?: ProviderMetadata;
}
| DataUIMessageChunk
| {
diff --git a/packages/ai/src/ui/call-completion-api.ts b/packages/ai/src/ui/call-completion-api.ts
index 2af1196e1df9..8174dac67ac2 100644
--- a/packages/ai/src/ui/call-completion-api.ts
+++ b/packages/ai/src/ui/call-completion-api.ts
@@ -1,10 +1,16 @@
-import { parseJsonEventStream, ParseResult } from '@ai-sdk/provider-utils';
+import {
+ parseJsonEventStream,
+ ParseResult,
+ withUserAgentSuffix,
+ getRuntimeEnvironmentUserAgent,
+} from '@ai-sdk/provider-utils';
import {
UIMessageChunk,
uiMessageChunkSchema,
} from '../ui-message-stream/ui-message-chunks';
import { consumeStream } from '../util/consume-stream';
import { processTextStream } from './process-text-stream';
+import { VERSION } from '../version';
// use function to allow for mocking in tests:
const getOriginalFetch = () => fetch;
@@ -55,10 +61,14 @@ export async function callCompletionApi({
...body,
}),
credentials,
- headers: {
- 'Content-Type': 'application/json',
- ...headers,
- },
+ headers: withUserAgentSuffix(
+ {
+ 'Content-Type': 'application/json',
+ ...headers,
+ },
+ `ai-sdk/${VERSION}`,
+ getRuntimeEnvironmentUserAgent(),
+ ),
signal: abortController.signal,
}).catch(err => {
throw err;
diff --git a/packages/ai/src/ui/chat.test.ts b/packages/ai/src/ui/chat.test.ts
index 6d95f9c7134e..81afc12e6707 100644
--- a/packages/ai/src/ui/chat.test.ts
+++ b/packages/ai/src/ui/chat.test.ts
@@ -1,8 +1,8 @@
import {
createTestServer,
- mockId,
TestResponseController,
-} from '@ai-sdk/provider-utils/test';
+} from '@ai-sdk/test-server/with-vitest';
+import { mockId } from '@ai-sdk/provider-utils/test';
import { createResolvablePromise } from '../util/create-resolvable-promise';
import { AbstractChat, ChatInit, ChatState, ChatStatus } from './chat';
import { UIMessage } from './ui-messages';
@@ -1742,6 +1742,149 @@ describe('Chat', () => {
`);
});
+ it('should send message when a tool error result is submitted', async () => {
+ server.urls['http://localhost:3000/api/chat'].response = [
+ {
+ type: 'stream-chunks',
+ chunks: [
+ formatChunk({ type: 'start' }),
+ formatChunk({ type: 'start-step' }),
+ formatChunk({
+ type: 'tool-input-available',
+ toolCallId: 'tool-call-0',
+ toolName: 'test-tool',
+ input: { testArg: 'test-value' },
+ }),
+ formatChunk({ type: 'finish-step' }),
+ formatChunk({ type: 'finish' }),
+ ],
+ },
+ {
+ type: 'stream-chunks',
+ chunks: [
+ formatChunk({ type: 'start' }),
+ formatChunk({ type: 'start-step' }),
+ formatChunk({ type: 'finish-step' }),
+ formatChunk({ type: 'finish' }),
+ ],
+ },
+ ];
+
+ let callCount = 0;
+ const onFinishPromise = createResolvablePromise();
+
+ const chat = new TestChat({
+ id: '123',
+ generateId: mockId(),
+ transport: new DefaultChatTransport({
+ api: 'http://localhost:3000/api/chat',
+ }),
+ sendAutomaticallyWhen: lastAssistantMessageIsCompleteWithToolCalls,
+ onFinish: () => {
+ callCount++;
+ if (callCount === 2) {
+ onFinishPromise.resolve();
+ }
+ },
+ });
+
+ await chat.sendMessage({
+ text: 'Hello, world!',
+ });
+
+ // user submits the tool result
+ await chat.addToolResult({
+ state: 'output-error',
+ tool: 'test-tool',
+ toolCallId: 'tool-call-0',
+ errorText: 'test-error',
+ });
+
+ // UI should show the tool result
+ expect(chat.messages).toMatchInlineSnapshot(`
+ [
+ {
+ "id": "id-0",
+ "metadata": undefined,
+ "parts": [
+ {
+ "text": "Hello, world!",
+ "type": "text",
+ },
+ ],
+ "role": "user",
+ },
+ {
+ "id": "id-1",
+ "metadata": undefined,
+ "parts": [
+ {
+ "type": "step-start",
+ },
+ {
+ "errorText": "test-error",
+ "input": {
+ "testArg": "test-value",
+ },
+ "output": undefined,
+ "preliminary": undefined,
+ "providerExecuted": undefined,
+ "rawInput": undefined,
+ "state": "output-error",
+ "toolCallId": "tool-call-0",
+ "type": "tool-test-tool",
+ },
+ ],
+ "role": "assistant",
+ },
+ ]
+ `);
+
+ await onFinishPromise.promise;
+
+ // 2nd call should happen after the stream is finished
+ expect(server.calls.length).toBe(2);
+
+ // check details of the 2nd call
+ expect(await server.calls[1].requestBodyJson).toMatchInlineSnapshot(`
+ {
+ "id": "123",
+ "messageId": "id-1",
+ "messages": [
+ {
+ "id": "id-0",
+ "parts": [
+ {
+ "text": "Hello, world!",
+ "type": "text",
+ },
+ ],
+ "role": "user",
+ },
+ {
+ "id": "id-1",
+ "parts": [
+ {
+ "type": "step-start",
+ },
+ {
+ "errorText": "test-error",
+ "input": {
+ "testArg": "test-value",
+ },
+ "state": "output-error",
+ "toolCallId": "tool-call-0",
+ "type": "tool-test-tool",
+ },
+ ],
+ "role": "assistant",
+ },
+ ],
+ "trigger": "submit-message",
+ }
+ `);
+ });
+
it('should send message when a dynamic tool result is submitted', async () => {
server.urls['http://localhost:3000/api/chat'].response = [
{
@@ -1795,6 +1938,7 @@ describe('Chat', () => {
// user submits the tool result
await chat.addToolResult({
+ state: 'output-available',
tool: 'test-tool',
toolCallId: 'tool-call-0',
output: 'test-result',
diff --git a/packages/ai/src/ui/chat.ts b/packages/ai/src/ui/chat.ts
index f1f94192211d..65831e407219 100644
--- a/packages/ai/src/ui/chat.ts
+++ b/packages/ai/src/ui/chat.ts
@@ -18,6 +18,7 @@ import {
import {
InferUIMessageToolCall,
isToolOrDynamicToolUIPart,
+ ToolUIPart,
type DataUIPart,
type FileUIPart,
type InferUIMessageData,
@@ -414,14 +415,26 @@ export abstract class AbstractChat {
};
addToolResult = async >({
+ state = 'output-available',
tool,
toolCallId,
output,
- }: {
- tool: TOOL;
- toolCallId: string;
- output: InferUIMessageTools[TOOL]['output'];
- }) =>
+ errorText,
+ }:
+ | {
+ state?: 'output-available';
+ tool: TOOL;
+ toolCallId: string;
+ output: InferUIMessageTools[TOOL]['output'];
+ errorText?: never;
+ }
+ | {
+ state: 'output-error';
+ tool: TOOL;
+ toolCallId: string;
+ output?: never;
+ errorText: string;
+ }) =>
this.jobExecutor.run(async () => {
const messages = this.state.messages;
const lastMessage = messages[messages.length - 1];
@@ -430,7 +443,7 @@ export abstract class AbstractChat {
...lastMessage,
parts: lastMessage.parts.map(part =>
isToolOrDynamicToolUIPart(part) && part.toolCallId === toolCallId
- ? { ...part, state: 'output-available', output }
+ ? { ...part, state, output, errorText }
: part,
),
});
@@ -440,12 +453,12 @@ export abstract class AbstractChat {
this.activeResponse.state.message.parts =
this.activeResponse.state.message.parts.map(part =>
isToolOrDynamicToolUIPart(part) && part.toolCallId === toolCallId
- ? {
+ ? ({
...part,
- state: 'output-available',
+ state,
output,
- errorText: undefined,
- }
+ errorText,
+ } as ToolUIPart>)
: part,
);
}
diff --git a/packages/ai/src/ui/http-chat-transport.test.ts b/packages/ai/src/ui/http-chat-transport.test.ts
index 031f537baff9..820dd4d0fc7c 100644
--- a/packages/ai/src/ui/http-chat-transport.test.ts
+++ b/packages/ai/src/ui/http-chat-transport.test.ts
@@ -1,4 +1,4 @@
-import { createTestServer } from '@ai-sdk/provider-utils/test';
+import { createTestServer } from '@ai-sdk/test-server/with-vitest';
import { UIMessageChunk } from '../ui-message-stream/ui-message-chunks';
import {
HttpChatTransport,
@@ -148,6 +148,7 @@ describe('HttpChatTransport', () => {
expect(server.calls[0].requestHeaders['x-test-header']).toBe(
'test-value',
);
+ expect(server.calls[0].requestUserAgent).toContain('ai-sdk/');
});
it('should include headers in the request when a function is provided', async () => {
@@ -178,6 +179,7 @@ describe('HttpChatTransport', () => {
expect(server.calls[0].requestHeaders['x-test-header']).toBe(
'test-value-fn',
);
+ expect(server.calls[0].requestUserAgent).toContain('ai-sdk/');
});
});
});
diff --git a/packages/ai/src/ui/http-chat-transport.ts b/packages/ai/src/ui/http-chat-transport.ts
index b0ed429eb5c6..d304e4f794cf 100644
--- a/packages/ai/src/ui/http-chat-transport.ts
+++ b/packages/ai/src/ui/http-chat-transport.ts
@@ -1,7 +1,14 @@
-import { FetchFunction, Resolvable, resolve } from '@ai-sdk/provider-utils';
+import {
+ FetchFunction,
+ Resolvable,
+ resolve,
+ withUserAgentSuffix,
+ getRuntimeEnvironmentUserAgent,
+} from '@ai-sdk/provider-utils';
import { UIMessageChunk } from '../ui-message-stream/ui-message-chunks';
import { ChatTransport } from './chat-transport';
import { UIMessage } from './ui-messages';
+import { VERSION } from '../version';
export type PrepareSendMessagesRequest = (
options: {
@@ -188,10 +195,14 @@ export abstract class HttpChatTransport
const response = await fetch(api, {
method: 'POST',
- headers: {
- 'Content-Type': 'application/json',
- ...headers,
- },
+ headers: withUserAgentSuffix(
+ {
+ 'Content-Type': 'application/json',
+ ...headers,
+ },
+ `ai-sdk/${VERSION}`,
+ getRuntimeEnvironmentUserAgent(),
+ ),
body: JSON.stringify(body),
credentials,
signal: abortSignal,
@@ -238,7 +249,11 @@ export abstract class HttpChatTransport
const response = await fetch(api, {
method: 'GET',
- headers,
+ headers: withUserAgentSuffix(
+ headers,
+ `ai-sdk/${VERSION}`,
+ getRuntimeEnvironmentUserAgent(),
+ ),
credentials,
});
diff --git a/packages/ai/src/ui/index.ts b/packages/ai/src/ui/index.ts
index 90eeffd5d51d..2bf6e0a1e13f 100644
--- a/packages/ai/src/ui/index.ts
+++ b/packages/ai/src/ui/index.ts
@@ -55,4 +55,8 @@ export {
type CompletionRequestOptions,
type UseCompletionOptions,
} from './use-completion';
-export { validateUIMessages } from './validate-ui-messages';
+export {
+ validateUIMessages,
+ safeValidateUIMessages,
+ type SafeValidateUIMessagesResult,
+} from './validate-ui-messages';
diff --git a/packages/ai/src/ui/last-assistant-message-is-complete-with-tool-calls.test.ts b/packages/ai/src/ui/last-assistant-message-is-complete-with-tool-calls.test.ts
index 526ab4172664..cf9b759099d3 100644
--- a/packages/ai/src/ui/last-assistant-message-is-complete-with-tool-calls.test.ts
+++ b/packages/ai/src/ui/last-assistant-message-is-complete-with-tool-calls.test.ts
@@ -2,7 +2,7 @@ import { lastAssistantMessageIsCompleteWithToolCalls } from './last-assistant-me
import { describe, it, expect } from 'vitest';
describe('lastAssistantMessageIsCompleteWithToolCalls', () => {
- it('should return false if the last step of a multi-step sequency only has text', () => {
+ it('should return false if the last step of a multi-step sequence only has text', () => {
expect(
lastAssistantMessageIsCompleteWithToolCalls({
messages: [
@@ -61,6 +61,36 @@ describe('lastAssistantMessageIsCompleteWithToolCalls', () => {
).toBe(true);
});
+ it('should return true when the tool has a output-error state', () => {
+ expect(
+ lastAssistantMessageIsCompleteWithToolCalls({
+ messages: [
+ {
+ id: '1',
+ role: 'assistant',
+ parts: [
+ { type: 'step-start' },
+ {
+ type: 'tool-getWeatherInformation',
+ toolCallId: 'call_6iy0GxZ9R4VDI5MKohXxV48y',
+ state: 'output-error',
+ input: {
+ city: 'New York',
+ },
+ errorText: 'Unable to get weather information',
+ },
+ {
+ type: 'text',
+ text: 'The current weather in New York is windy.',
+ state: 'done',
+ },
+ ],
+ },
+ ],
+ }),
+ ).toBe(true);
+ });
+
it('should return true when dynamic tool call is complete', () => {
expect(
lastAssistantMessageIsCompleteWithToolCalls({
@@ -137,7 +167,7 @@ describe('lastAssistantMessageIsCompleteWithToolCalls', () => {
).toBe(false);
});
- it('should return false when dynamic tool call has an error', () => {
+ it('should return true when dynamic tool call has an error', () => {
expect(
lastAssistantMessageIsCompleteWithToolCalls({
messages: [
@@ -160,7 +190,7 @@ describe('lastAssistantMessageIsCompleteWithToolCalls', () => {
},
],
}),
- ).toBe(false);
+ ).toBe(true);
});
it('should return true when mixing regular and dynamic tool calls and all are complete', () => {
diff --git a/packages/ai/src/ui/last-assistant-message-is-complete-with-tool-calls.ts b/packages/ai/src/ui/last-assistant-message-is-complete-with-tool-calls.ts
index da7b2c386a2f..613430ac1a20 100644
--- a/packages/ai/src/ui/last-assistant-message-is-complete-with-tool-calls.ts
+++ b/packages/ai/src/ui/last-assistant-message-is-complete-with-tool-calls.ts
@@ -1,4 +1,4 @@
-import { isToolOrDynamicToolUIPart, UIMessage } from './ui-messages';
+import { isToolOrDynamicToolUIPart, type UIMessage } from './ui-messages';
/**
Check if the message is an assistant message with completed tool calls.
@@ -30,6 +30,9 @@ export function lastAssistantMessageIsCompleteWithToolCalls({
return (
lastStepToolInvocations.length > 0 &&
- lastStepToolInvocations.every(part => part.state === 'output-available')
+ lastStepToolInvocations.every(
+ part =>
+ part.state === 'output-available' || part.state === 'output-error',
+ )
);
}
diff --git a/packages/ai/src/ui/validate-ui-messages.test.ts b/packages/ai/src/ui/validate-ui-messages.test.ts
index c413768bf93c..00ecd018d23b 100644
--- a/packages/ai/src/ui/validate-ui-messages.test.ts
+++ b/packages/ai/src/ui/validate-ui-messages.test.ts
@@ -1,6 +1,9 @@
import { z } from 'zod/v4';
import { InferUITool, UIMessage } from './ui-messages';
-import { validateUIMessages } from './validate-ui-messages';
+import {
+ safeValidateUIMessages,
+ validateUIMessages,
+} from './validate-ui-messages';
import { describe, it, expect, expectTypeOf } from 'vitest';
describe('validateUIMessages', () => {
@@ -678,6 +681,7 @@ describe('validateUIMessages', () => {
toolCallId: '1',
state: 'input-streaming',
input: { foo: 'bar' },
+ providerExecuted: true,
},
],
},
@@ -695,6 +699,7 @@ describe('validateUIMessages', () => {
"input": {
"foo": "bar",
},
+ "providerExecuted": true,
"state": "input-streaming",
"toolCallId": "1",
"type": "tool-foo",
@@ -718,6 +723,7 @@ describe('validateUIMessages', () => {
toolCallId: '1',
state: 'input-available',
input: { foo: 'bar' },
+ providerExecuted: true,
},
],
},
@@ -735,6 +741,7 @@ describe('validateUIMessages', () => {
"input": {
"foo": "bar",
},
+ "providerExecuted": true,
"state": "input-available",
"toolCallId": "1",
"type": "tool-foo",
@@ -759,6 +766,7 @@ describe('validateUIMessages', () => {
state: 'output-available',
input: { foo: 'bar' },
output: { result: 'success' },
+ providerExecuted: true,
},
],
},
@@ -779,6 +787,7 @@ describe('validateUIMessages', () => {
"output": {
"result": "success",
},
+ "providerExecuted": true,
"state": "output-available",
"toolCallId": "1",
"type": "tool-foo",
@@ -803,6 +812,7 @@ describe('validateUIMessages', () => {
state: 'output-error',
input: { foo: 'bar' },
errorText: 'Tool execution failed',
+ providerExecuted: true,
},
],
},
@@ -821,6 +831,7 @@ describe('validateUIMessages', () => {
"input": {
"foo": "bar",
},
+ "providerExecuted": true,
"state": "output-error",
"toolCallId": "1",
"type": "tool-foo",
@@ -844,6 +855,7 @@ describe('validateUIMessages', () => {
toolCallId: '1',
state: 'input-available',
input: { foo: 'bar' },
+ providerExecuted: true,
},
],
},
@@ -855,23 +867,24 @@ describe('validateUIMessages', () => {
expectTypeOf(messages).toEqualTypeOf>();
expect(messages).toMatchInlineSnapshot(`
- [
- {
- "id": "1",
- "parts": [
- {
- "input": {
- "foo": "bar",
+ [
+ {
+ "id": "1",
+ "parts": [
+ {
+ "input": {
+ "foo": "bar",
+ },
+ "providerExecuted": true,
+ "state": "input-available",
+ "toolCallId": "1",
+ "type": "tool-foo",
},
- "state": "input-available",
- "toolCallId": "1",
- "type": "tool-foo",
- },
- ],
- "role": "assistant",
- },
- ]
- `);
+ ],
+ "role": "assistant",
+ },
+ ]
+ `);
});
it('should validate tool input and output when state is output-available', async () => {
@@ -933,6 +946,7 @@ describe('validateUIMessages', () => {
state: 'output-error',
input: { foo: 'bar' },
errorText: 'Tool execution failed',
+ providerExecuted: true,
},
],
},
@@ -953,6 +967,7 @@ describe('validateUIMessages', () => {
"input": {
"foo": "bar",
},
+ "providerExecuted": true,
"state": "output-error",
"toolCallId": "1",
"type": "tool-foo",
@@ -977,6 +992,7 @@ describe('validateUIMessages', () => {
toolCallId: '1',
state: 'input-available',
input: { foo: 'bar' },
+ providerExecuted: true,
},
],
},
@@ -1004,6 +1020,7 @@ describe('validateUIMessages', () => {
toolCallId: '1',
state: 'input-available',
input: { foo: 123 }, // wrong type
+ providerExecuted: true,
},
],
},
@@ -1030,6 +1047,7 @@ describe('validateUIMessages', () => {
type: 'tool-foo',
toolCallId: '1',
state: 'output-available',
+ providerExecuted: true,
input: { foo: 'bar' },
output: { result: 123 }, // wrong type
},
@@ -1058,6 +1076,7 @@ describe('validateUIMessages', () => {
toolCallId: '1',
state: 'input-streaming',
input: { foo: 123 }, // wrong type but should not be validated
+ providerExecuted: true,
},
],
},
@@ -1077,6 +1096,7 @@ describe('validateUIMessages', () => {
"input": {
"foo": 123,
},
+ "providerExecuted": true,
"state": "input-streaming",
"toolCallId": "1",
"type": "tool-foo",
@@ -1089,3 +1109,135 @@ describe('validateUIMessages', () => {
});
});
});
+
+export function expectToBe(
+ value: boolean,
+ expected: T,
+): asserts value is T {
+ expect(value).toBe(expected);
+}
+
+describe('safeValidateUIMessages', () => {
+ it('should return success result for valid messages', async () => {
+ const result = await safeValidateUIMessages({
+ messages: [
+ {
+ id: '1',
+ role: 'user',
+ parts: [{ type: 'text', text: 'Hello, world!' }],
+ },
+ ],
+ });
+
+ expectToBe(result.success, true);
+ expect(result.data).toMatchInlineSnapshot(`
+ [
+ {
+ "id": "1",
+ "parts": [
+ {
+ "text": "Hello, world!",
+ "type": "text",
+ },
+ ],
+ "role": "user",
+ },
+ ]
+ `);
+ });
+
+ it('should return failure result when messages parameter is null', async () => {
+ const result = await safeValidateUIMessages({
+ messages: null,
+ });
+
+ expectToBe(result.success, false);
+ expect(result.error.name).toBe('AI_InvalidArgumentError');
+ expect(result.error.message).toBe(
+ 'Invalid argument for parameter messages: messages parameter must be provided',
+ );
+ });
+
+ it('should return failure result when metadata validation fails', async () => {
+ const result = await safeValidateUIMessages>({
+ messages: [
+ {
+ id: '1',
+ role: 'user',
+ metadata: { foo: 123 },
+ parts: [{ type: 'text', text: 'Hello, world!' }],
+ },
+ ],
+ metadataSchema: z.object({ foo: z.string() }),
+ });
+
+ expectToBe(result.success, false);
+ expect(result.error.name).toBe('AI_TypeValidationError');
+ expect(result.error.message).toContain('Type validation failed');
+ });
+
+ it('should return failure result when tool input validation fails', async () => {
+ const testTool = {
+ name: 'foo',
+ inputSchema: z.object({ foo: z.string() }),
+ outputSchema: z.object({ result: z.string() }),
+ };
+
+ const result = await safeValidateUIMessages({
+ messages: [
+ {
+ id: '1',
+ role: 'assistant',
+ parts: [
+ {
+ type: 'tool-foo',
+ toolCallId: '1',
+ state: 'input-available',
+ input: { foo: 123 },
+ providerExecuted: true,
+ },
+ ],
+ },
+ ],
+ tools: { foo: testTool },
+ });
+
+ expectToBe(result.success, false);
+ expect(result.error.name).toBe('AI_TypeValidationError');
+ expect(result.error.message).toContain('Type validation failed');
+ });
+
+ it('should return failure result when data schema is missing', async () => {
+ const result = await safeValidateUIMessages({
+ messages: [
+ {
+ id: '1',
+ role: 'assistant',
+ parts: [{ type: 'data-bar', data: { foo: 'bar' } }],
+ },
+ ],
+ dataSchemas: {
+ foo: z.object({ foo: z.string() }),
+ },
+ });
+
+ expectToBe(result.success, false);
+ expect(result.error.name).toBe('AI_TypeValidationError');
+ expect(result.error.message).toContain(
+ 'No data schema found for data part bar',
+ );
+ });
+
+ it('should return failure result for invalid message structure', async () => {
+ const result = await safeValidateUIMessages({
+ messages: [
+ {
+ role: 'user',
+ },
+ ],
+ });
+
+ expectToBe(result.success, false);
+ expect(result.error.name).toBe('AI_TypeValidationError');
+ });
+});
diff --git a/packages/ai/src/ui/validate-ui-messages.ts b/packages/ai/src/ui/validate-ui-messages.ts
index 131d4e9d43b2..c82a6e11c00b 100644
--- a/packages/ai/src/ui/validate-ui-messages.ts
+++ b/packages/ai/src/ui/validate-ui-messages.ts
@@ -113,6 +113,7 @@ const toolUIPartSchemas = [
type: z.string().startsWith('tool-'),
toolCallId: z.string(),
state: z.literal('input-streaming'),
+ providerExecuted: z.boolean().optional(),
input: z.unknown().optional(),
output: z.never().optional(),
errorText: z.never().optional(),
@@ -121,6 +122,7 @@ const toolUIPartSchemas = [
type: z.string().startsWith('tool-'),
toolCallId: z.string(),
state: z.literal('input-available'),
+ providerExecuted: z.boolean().optional(),
input: z.unknown(),
output: z.never().optional(),
errorText: z.never().optional(),
@@ -130,6 +132,7 @@ const toolUIPartSchemas = [
type: z.string().startsWith('tool-'),
toolCallId: z.string(),
state: z.literal('output-available'),
+ providerExecuted: z.boolean().optional(),
input: z.unknown(),
output: z.unknown(),
errorText: z.never().optional(),
@@ -140,6 +143,7 @@ const toolUIPartSchemas = [
type: z.string().startsWith('tool-'),
toolCallId: z.string(),
state: z.literal('output-error'),
+ providerExecuted: z.boolean().optional(),
input: z.unknown(),
output: z.never().optional(),
errorText: z.string(),
@@ -166,14 +170,22 @@ const uiMessageSchema = z.object({
),
});
+export type SafeValidateUIMessagesResult =
+ | {
+ success: true;
+ data: Array;
+ }
+ | {
+ success: false;
+ error: Error;
+ };
+
/**
- * Validates a list of UI messages.
- *
- * Metadata, data parts, and generic tool call structures are only validated if
- * the corresponding schemas are provided. Otherwise, they are assumed to be
- * valid.
+ * Validates a list of UI messages like `validateUIMessages`,
+ * but instead of throwing it returns `{ success: true, data }`
+ * or `{ success: false, error }`.
*/
-export async function validateUIMessages({
+export async function safeValidateUIMessages({
messages,
metadataSchema,
dataSchemas,
@@ -194,91 +206,153 @@ export async function validateUIMessages({
InferUIMessageTools[NAME]['output']
>;
};
-}): Promise> {
- if (messages == null) {
- throw new InvalidArgumentError({
- parameter: 'messages',
+}): Promise> {
+ try {
+ if (messages == null) {
+ return {
+ success: false,
+ error: new InvalidArgumentError({
+ parameter: 'messages',
+ value: messages,
+ message: 'messages parameter must be provided',
+ }),
+ };
+ }
+
+ const validatedMessages = await validateTypes({
value: messages,
- message: 'messages parameter must be provided',
+ schema: z.array(uiMessageSchema),
});
- }
-
- const validatedMessages = await validateTypes({
- value: messages,
- schema: z.array(uiMessageSchema),
- });
- if (metadataSchema) {
- for (const message of validatedMessages) {
- await validateTypes({
- value: message.metadata,
- schema: metadataSchema,
- });
+ if (metadataSchema) {
+ for (const message of validatedMessages) {
+ await validateTypes({
+ value: message.metadata,
+ schema: metadataSchema,
+ });
+ }
}
- }
- if (dataSchemas) {
- for (const message of validatedMessages) {
- const dataParts = message.parts.filter(part =>
- part.type.startsWith('data-'),
- ) as DataUIPart>[];
+ if (dataSchemas) {
+ for (const message of validatedMessages) {
+ const dataParts = message.parts.filter(part =>
+ part.type.startsWith('data-'),
+ ) as DataUIPart>[];
+
+ for (const dataPart of dataParts) {
+ const dataName = dataPart.type.slice(5);
+ const dataSchema = dataSchemas[dataName];
- for (const dataPart of dataParts) {
- const dataName = dataPart.type.slice(5);
- const dataSchema = dataSchemas[dataName];
+ if (!dataSchema) {
+ return {
+ success: false,
+ error: new TypeValidationError({
+ value: dataPart.data,
+ cause: `No data schema found for data part ${dataName}`,
+ }),
+ };
+ }
- if (!dataSchema) {
- throw new TypeValidationError({
+ await validateTypes({
value: dataPart.data,
- cause: `No data schema found for data part ${dataName}`,
+ schema: dataSchema,
});
}
-
- await validateTypes({
- value: dataPart.data,
- schema: dataSchema,
- });
}
}
- }
- if (tools) {
- for (const message of validatedMessages) {
- const toolParts = message.parts.filter(part =>
- part.type.startsWith('tool-'),
- ) as ToolUIPart>[];
+ if (tools) {
+ for (const message of validatedMessages) {
+ const toolParts = message.parts.filter(part =>
+ part.type.startsWith('tool-'),
+ ) as ToolUIPart>[];
- for (const toolPart of toolParts) {
- const toolName = toolPart.type.slice(5);
- const tool = tools[toolName];
+ for (const toolPart of toolParts) {
+ const toolName = toolPart.type.slice(5);
+ const tool = tools[toolName];
- if (!tool) {
- throw new TypeValidationError({
- value: toolPart.input,
- cause: `No tool schema found for tool part ${toolName}`,
- });
- }
+ if (!tool) {
+ return {
+ success: false,
+ error: new TypeValidationError({
+ value: toolPart.input,
+ cause: `No tool schema found for tool part ${toolName}`,
+ }),
+ };
+ }
- if (
- toolPart.state === 'input-available' ||
- toolPart.state === 'output-available' ||
- toolPart.state === 'output-error'
- ) {
- await validateTypes({
- value: toolPart.input,
- schema: tool.inputSchema,
- });
- }
+ if (
+ toolPart.state === 'input-available' ||
+ toolPart.state === 'output-available' ||
+ toolPart.state === 'output-error'
+ ) {
+ await validateTypes({
+ value: toolPart.input,
+ schema: tool.inputSchema,
+ });
+ }
- if (toolPart.state === 'output-available' && tool.outputSchema) {
- await validateTypes({
- value: toolPart.output,
- schema: tool.outputSchema,
- });
+ if (toolPart.state === 'output-available' && tool.outputSchema) {
+ await validateTypes({
+ value: toolPart.output,
+ schema: tool.outputSchema,
+ });
+ }
}
}
}
+
+ return {
+ success: true,
+ data: validatedMessages as Array,
+ };
+ } catch (error) {
+ const err = error as Error;
+
+ return {
+ success: false,
+ error: err,
+ };
}
+}
+
+/**
+ * Validates a list of UI messages.
+ *
+ * Metadata, data parts, and generic tool call structures are only validated if
+ * the corresponding schemas are provided. Otherwise, they are assumed to be
+ * valid.
+ */
+export async function validateUIMessages({
+ messages,
+ metadataSchema,
+ dataSchemas,
+ tools,
+}: {
+ messages: unknown;
+ metadataSchema?:
+ | Validator
+ | StandardSchemaV1;
+ dataSchemas?: {
+ [NAME in keyof InferUIMessageData & string]?:
+ | Validator[NAME]>
+ | StandardSchemaV1[NAME]>;
+ };
+ tools?: {
+ [NAME in keyof InferUIMessageTools & string]?: Tool<
+ InferUIMessageTools[NAME]['input'],
+ InferUIMessageTools[NAME]['output']
+ >;
+ };
+}): Promise> {
+ const response = await safeValidateUIMessages({
+ messages,
+ metadataSchema,
+ dataSchemas,
+ tools,
+ });
+
+ if (!response.success) throw response.error;
- return validatedMessages as Array;
+ return response.data;
}
diff --git a/packages/ai/src/util/detect-media-type.test.ts b/packages/ai/src/util/detect-media-type.test.ts
index f127ea819e89..0962f08aa47a 100644
--- a/packages/ai/src/util/detect-media-type.test.ts
+++ b/packages/ai/src/util/detect-media-type.test.ts
@@ -74,8 +74,26 @@ describe('detectMediaType', () => {
});
describe('WebP', () => {
- it('should detect WebP from bytes', () => {
- const webpBytes = new Uint8Array([0x52, 0x49, 0x46, 0x46, 0xff, 0xff]);
+ it('should detect WebP from bytes (positive webp image uint8)', () => {
+ // WebP format: RIFF + 4 bytes file size + WEBP
+ const webpBytes = new Uint8Array([
+ 0x52,
+ 0x49,
+ 0x46,
+ 0x46, // "RIFF"
+ 0x24,
+ 0x00,
+ 0x00,
+ 0x00, // file size (example: 36 bytes)
+ 0x57,
+ 0x45,
+ 0x42,
+ 0x50, // "WEBP"
+ 0x56,
+ 0x50,
+ 0x38,
+ 0x20, // VP8 chunk (additional WebP data)
+ ]);
expect(
detectMediaType({
data: webpBytes,
@@ -84,8 +102,27 @@ describe('detectMediaType', () => {
).toBe('image/webp');
});
- it('should detect WebP from base64', () => {
- const webpBase64 = 'UklGRgabc123'; // Base64 string starting with WebP signature
+ it('should detect WebP from base64 (positive webp image base64)', () => {
+ // WebP: RIFF + file size + WEBP encoded to base64
+ const webpBytes = new Uint8Array([
+ 0x52,
+ 0x49,
+ 0x46,
+ 0x46, // "RIFF"
+ 0x24,
+ 0x00,
+ 0x00,
+ 0x00, // file size
+ 0x57,
+ 0x45,
+ 0x42,
+ 0x50, // "WEBP"
+ 0x56,
+ 0x50,
+ 0x38,
+ 0x20, // VP8 chunk
+ ]);
+ const webpBase64 = convertUint8ArrayToBase64(webpBytes);
expect(
detectMediaType({
data: webpBase64,
@@ -93,6 +130,63 @@ describe('detectMediaType', () => {
}),
).toBe('image/webp');
});
+
+ it('should NOT detect RIFF audio as WebP from bytes (negative riff audio uint8)', () => {
+ // WAV format: RIFF + file size + WAVE (not WEBP)
+ const wavBytes = new Uint8Array([
+ 0x52,
+ 0x49,
+ 0x46,
+ 0x46, // "RIFF"
+ 0x24,
+ 0x00,
+ 0x00,
+ 0x00, // file size
+ 0x57,
+ 0x41,
+ 0x56,
+ 0x45, // "WAVE" (not "WEBP")
+ 0x66,
+ 0x6d,
+ 0x74,
+ 0x20, // fmt chunk
+ ]);
+ expect(
+ detectMediaType({
+ data: wavBytes,
+ signatures: imageMediaTypeSignatures,
+ }),
+ ).toBeUndefined(); // Should not detect as WebP
+ });
+
+ it('should NOT detect RIFF audio as WebP from base64 (negative riff audio base64)', () => {
+ // WAV format encoded to base64
+ const wavBytes = new Uint8Array([
+ 0x52,
+ 0x49,
+ 0x46,
+ 0x46, // "RIFF"
+ 0x24,
+ 0x00,
+ 0x00,
+ 0x00, // file size
+ 0x57,
+ 0x41,
+ 0x56,
+ 0x45, // "WAVE" (not "WEBP")
+ 0x66,
+ 0x6d,
+ 0x74,
+ 0x20, // fmt chunk
+ ]);
+ const wavBase64 = convertUint8ArrayToBase64(wavBytes);
+ expect(
+ detectMediaType({
+ data: wavBase64,
+ signatures: imageMediaTypeSignatures,
+ }),
+ ).toBeUndefined(); // Should not detect as WebP
+ });
});
describe('BMP', () => {
@@ -107,10 +201,10 @@ describe('detectMediaType', () => {
});
it('should detect BMP from base64', () => {
- const bmpBase64 = 'Qkabc123'; // Base64 string starting with BMP signature
+ const bmpBytes = new Uint8Array([0x42, 0x4d, 0xff, 0xff]);
expect(
detectMediaType({
- data: bmpBase64,
+ data: convertUint8ArrayToBase64(bmpBytes),
signatures: imageMediaTypeSignatures,
}),
).toBe('image/bmp');
@@ -306,8 +400,46 @@ describe('detectMediaType', () => {
});
describe('WAV', () => {
+ // WebP format: RIFF + 4 bytes file size + WEBP
+ const webpBytes = new Uint8Array([
+ 0x52,
+ 0x49,
+ 0x46,
+ 0x46, // "RIFF"
+ 0x24,
+ 0x00,
+ 0x00,
+ 0x00, // file size (example: 36 bytes)
+ 0x57,
+ 0x45,
+ 0x42,
+ 0x50, // "WEBP"
+ 0x56,
+ 0x50,
+ 0x38,
+ 0x20, // VP8 chunk (additional WebP data)
+ ]);
+
+ const wavBytes = new Uint8Array([
+ 0x52,
+ 0x49,
+ 0x46,
+ 0x46, // "RIFF"
+ 0x24,
+ 0x00,
+ 0x00,
+ 0x00, // file size (example: 36 bytes)
+ 0x57,
+ 0x41,
+ 0x56,
+ 0x45, // "WAVE" (not "WEBP")
+ 0x66,
+ 0x6d,
+ 0x74,
+ 0x20, // fmt chunk
+ ]);
+
it('should detect WAV from bytes', () => {
- const wavBytes = new Uint8Array([0x52, 0x49, 0x46, 0x46]);
expect(
detectMediaType({
data: wavBytes,
@@ -317,14 +449,31 @@ describe('detectMediaType', () => {
});
it('should detect WAV from base64', () => {
- const wavBase64 = 'UklGRiQ='; // Base64 string starting with WAV signature
expect(
detectMediaType({
- data: wavBase64,
+ data: convertUint8ArrayToBase64(wavBytes),
signatures: audioMediaTypeSignatures,
}),
).toBe('audio/wav');
});
+
+ it('should NOT detect WebP as WAV from bytes (negative webp image uint8)', () => {
+ expect(
+ detectMediaType({
+ data: webpBytes,
+ signatures: audioMediaTypeSignatures,
+ }),
+ ).toBeUndefined(); // Should not detect as WAV
+ });
+
+ it('should NOT detect WebP as WAV from base64 (negative webp image base64)', () => {
+ expect(
+ detectMediaType({
+ data: convertUint8ArrayToBase64(webpBytes),
+ signatures: audioMediaTypeSignatures,
+ }),
+ ).toBeUndefined(); // Should not detect as WAV
+ });
});
describe('OGG', () => {
@@ -383,10 +532,10 @@ describe('detectMediaType', () => {
});
it('should detect AAC from base64', () => {
- const aacBase64 = 'QBUA'; // Base64 string starting with AAC signature
+ const aacBytes = new Uint8Array([0x40, 0x15, 0x00, 0x00]);
expect(
detectMediaType({
- data: aacBase64,
+ data: convertUint8ArrayToBase64(aacBytes),
signatures: audioMediaTypeSignatures,
}),
).toBe('audio/aac');
diff --git a/packages/ai/src/util/detect-media-type.ts b/packages/ai/src/util/detect-media-type.ts
index d36b2393411f..49a3ed237f4a 100644
--- a/packages/ai/src/util/detect-media-type.ts
+++ b/packages/ai/src/util/detect-media-type.ts
@@ -3,52 +3,56 @@ import { convertBase64ToUint8Array } from '@ai-sdk/provider-utils';
export const imageMediaTypeSignatures = [
{
mediaType: 'image/gif' as const,
- bytesPrefix: [0x47, 0x49, 0x46],
- base64Prefix: 'R0lG',
+ bytesPrefix: [0x47, 0x49, 0x46], // GIF
},
{
mediaType: 'image/png' as const,
- bytesPrefix: [0x89, 0x50, 0x4e, 0x47],
- base64Prefix: 'iVBORw',
+ bytesPrefix: [0x89, 0x50, 0x4e, 0x47], // PNG
},
{
mediaType: 'image/jpeg' as const,
- bytesPrefix: [0xff, 0xd8],
- base64Prefix: '/9j/',
+ bytesPrefix: [0xff, 0xd8], // JPEG
},
{
mediaType: 'image/webp' as const,
- bytesPrefix: [0x52, 0x49, 0x46, 0x46],
- base64Prefix: 'UklGRg',
+ bytesPrefix: [
+ 0x52,
+ 0x49,
+ 0x46,
+ 0x46, // "RIFF"
+ null,
+ null,
+ null,
+ null, // file size (variable)
+ 0x57,
+ 0x45,
+ 0x42,
+ 0x50, // "WEBP"
+ ],
},
{
mediaType: 'image/bmp' as const,
bytesPrefix: [0x42, 0x4d],
- base64Prefix: 'Qk',
},
{
mediaType: 'image/tiff' as const,
bytesPrefix: [0x49, 0x49, 0x2a, 0x00],
- base64Prefix: 'SUkqAA',
},
{
mediaType: 'image/tiff' as const,
bytesPrefix: [0x4d, 0x4d, 0x00, 0x2a],
- base64Prefix: 'TU0AKg',
},
{
mediaType: 'image/avif' as const,
bytesPrefix: [
0x00, 0x00, 0x00, 0x20, 0x66, 0x74, 0x79, 0x70, 0x61, 0x76, 0x69, 0x66,
],
- base64Prefix: 'AAAAIGZ0eXBhdmlm',
},
{
mediaType: 'image/heic' as const,
bytesPrefix: [
0x00, 0x00, 0x00, 0x20, 0x66, 0x74, 0x79, 0x70, 0x68, 0x65, 0x69, 0x63,
],
- base64Prefix: 'AAAAIGZ0eXBoZWlj',
},
] as const;
@@ -56,62 +60,63 @@ export const audioMediaTypeSignatures = [
{
mediaType: 'audio/mpeg' as const,
bytesPrefix: [0xff, 0xfb],
- base64Prefix: '//s=',
},
{
mediaType: 'audio/mpeg' as const,
bytesPrefix: [0xff, 0xfa],
- base64Prefix: '//o=',
},
{
mediaType: 'audio/mpeg' as const,
bytesPrefix: [0xff, 0xf3],
- base64Prefix: '//M=',
},
{
mediaType: 'audio/mpeg' as const,
bytesPrefix: [0xff, 0xf2],
- base64Prefix: '//I=',
},
{
mediaType: 'audio/mpeg' as const,
bytesPrefix: [0xff, 0xe3],
- base64Prefix: '/+M=',
},
{
mediaType: 'audio/mpeg' as const,
bytesPrefix: [0xff, 0xe2],
- base64Prefix: '/+I=',
},
{
mediaType: 'audio/wav' as const,
- bytesPrefix: [0x52, 0x49, 0x46, 0x46],
- base64Prefix: 'UklGR',
+ bytesPrefix: [
+ 0x52, // R
+ 0x49, // I
+ 0x46, // F
+ 0x46, // F
+ null,
+ null,
+ null,
+ null,
+ 0x57, // W
+ 0x41, // A
+ 0x56, // V
+ 0x45, // E
+ ],
},
{
mediaType: 'audio/ogg' as const,
bytesPrefix: [0x4f, 0x67, 0x67, 0x53],
- base64Prefix: 'T2dnUw',
},
{
mediaType: 'audio/flac' as const,
bytesPrefix: [0x66, 0x4c, 0x61, 0x43],
- base64Prefix: 'ZkxhQw',
},
{
mediaType: 'audio/aac' as const,
bytesPrefix: [0x40, 0x15, 0x00, 0x00],
- base64Prefix: 'QBUA',
},
{
mediaType: 'audio/mp4' as const,
bytesPrefix: [0x66, 0x74, 0x79, 0x70],
- base64Prefix: 'ZnR5cA',
},
{
mediaType: 'audio/webm',
bytesPrefix: [0x1a, 0x45, 0xdf, 0xa3],
- base64Prefix: 'GkXf',
},
] as const;
@@ -156,14 +161,20 @@ export function detectMediaType({
}): (typeof signatures)[number]['mediaType'] | undefined {
const processedData = stripID3TagsIfPresent(data);
+ // Convert the first ~18 bytes (24 base64 chars) for consistent detection logic:
+ const bytes =
+ typeof processedData === 'string'
+ ? convertBase64ToUint8Array(
+ processedData.substring(0, Math.min(processedData.length, 24)),
+ )
+ : processedData;
+
for (const signature of signatures) {
if (
- typeof processedData === 'string'
- ? processedData.startsWith(signature.base64Prefix)
- : processedData.length >= signature.bytesPrefix.length &&
- signature.bytesPrefix.every(
- (byte, index) => processedData[index] === byte,
- )
+ bytes.length >= signature.bytesPrefix.length &&
+ signature.bytesPrefix.every(
+ (byte, index) => byte === null || bytes[index] === byte,
+ )
) {
return signature.mediaType;
}
diff --git a/packages/ai/src/util/download/download.test.ts b/packages/ai/src/util/download/download.test.ts
index 8f483ed88e79..d239b4be5acf 100644
--- a/packages/ai/src/util/download/download.test.ts
+++ b/packages/ai/src/util/download/download.test.ts
@@ -1,4 +1,4 @@
-import { createTestServer } from '@ai-sdk/provider-utils/test';
+import { createTestServer } from '@ai-sdk/test-server/with-vitest';
import { download } from './download';
import { DownloadError } from './download-error';
import { describe, it, expect } from 'vitest';
@@ -26,6 +26,9 @@ describe('download', () => {
expect(result).not.toBeNull();
expect(result!.data).toEqual(expectedBytes);
expect(result!.mediaType).toBe('application/octet-stream');
+
+ // UA header assertion
+ expect(server.calls[0].requestUserAgent).toContain('ai-sdk/');
});
it('should throw DownloadError when response is not ok', async () => {
diff --git a/packages/ai/src/util/download/download.ts b/packages/ai/src/util/download/download.ts
index 84ae6d6db0de..57b3ca2bcd7b 100644
--- a/packages/ai/src/util/download/download.ts
+++ b/packages/ai/src/util/download/download.ts
@@ -1,4 +1,9 @@
import { DownloadError } from './download-error';
+import {
+ withUserAgentSuffix,
+ getRuntimeEnvironmentUserAgent,
+} from '@ai-sdk/provider-utils';
+import { VERSION } from '../../version';
/**
* Download a file from a URL.
@@ -11,7 +16,13 @@ import { DownloadError } from './download-error';
export const download = async ({ url }: { url: URL }) => {
const urlText = url.toString();
try {
- const response = await fetch(urlText);
+ const response = await fetch(urlText, {
+ headers: withUserAgentSuffix(
+ {},
+ `ai-sdk/${VERSION}`,
+ getRuntimeEnvironmentUserAgent(),
+ ),
+ });
if (!response.ok) {
throw new DownloadError({
diff --git a/packages/ai/src/version.ts b/packages/ai/src/version.ts
new file mode 100644
index 000000000000..8fda877d6d33
--- /dev/null
+++ b/packages/ai/src/version.ts
@@ -0,0 +1,5 @@
+declare const __PACKAGE_VERSION__: string | undefined;
+export const VERSION: string =
+ typeof __PACKAGE_VERSION__ !== 'undefined'
+ ? __PACKAGE_VERSION__
+ : '0.0.0-test';
diff --git a/packages/ai/test/index.ts b/packages/ai/test/index.ts
index 4068a8a2814f..fe6aedf7bbce 100644
--- a/packages/ai/test/index.ts
+++ b/packages/ai/test/index.ts
@@ -4,10 +4,10 @@ export {
convertReadableStreamToArray,
mockId,
} from '@ai-sdk/provider-utils/test';
-export { MockEmbeddingModelV2 } from '../src/test/mock-embedding-model-v2';
-export { MockImageModelV2 } from '../src/test/mock-image-model-v2';
+export { MockEmbeddingModelV3 } from '../src/test/mock-embedding-model-v3';
+export { MockImageModelV3 } from '../src/test/mock-image-model-v3';
export { MockLanguageModelV2 } from '../src/test/mock-language-model-v2';
-export { MockProviderV2 } from '../src/test/mock-provider-v2';
+export { MockProviderV3 } from '../src/test/mock-provider-v3';
export { MockSpeechModelV2 } from '../src/test/mock-speech-model-v2';
export { MockTranscriptionModelV2 } from '../src/test/mock-transcription-model-v2';
export { mockValues } from '../src/test/mock-values';
diff --git a/packages/ai/tsconfig.json b/packages/ai/tsconfig.json
index 7ac521e7e88f..7dedfdb70326 100644
--- a/packages/ai/tsconfig.json
+++ b/packages/ai/tsconfig.json
@@ -27,6 +27,9 @@
},
{
"path": "../gateway"
+ },
+ {
+ "path": "../test-server"
}
]
}
diff --git a/packages/ai/tsup.config.ts b/packages/ai/tsup.config.ts
index 6efc434cff45..40b286074930 100644
--- a/packages/ai/tsup.config.ts
+++ b/packages/ai/tsup.config.ts
@@ -10,6 +10,12 @@ export default defineConfig([
sourcemap: true,
target: 'es2018',
platform: 'node',
+ define: {
+ __PACKAGE_VERSION__: JSON.stringify(
+ (await import('./package.json', { with: { type: 'json' } })).default
+ .version,
+ ),
+ },
},
// Internal APIs
{
@@ -21,6 +27,12 @@ export default defineConfig([
sourcemap: true,
target: 'es2018',
platform: 'node',
+ define: {
+ __PACKAGE_VERSION__: JSON.stringify(
+ (await import('./package.json', { with: { type: 'json' } })).default
+ .version,
+ ),
+ },
},
// Test utilities
{
@@ -43,6 +55,12 @@ export default defineConfig([
// Allow BigInt in tests
target: 'es2020',
platform: 'node',
+ define: {
+ __PACKAGE_VERSION__: JSON.stringify(
+ (await import('./package.json', { with: { type: 'json' } })).default
+ .version,
+ ),
+ },
},
// MCP stdio
{
@@ -54,5 +72,11 @@ export default defineConfig([
sourcemap: true,
target: 'es2018',
platform: 'node',
+ define: {
+ __PACKAGE_VERSION__: JSON.stringify(
+ (await import('./package.json', { with: { type: 'json' } })).default
+ .version,
+ ),
+ },
},
]);
diff --git a/packages/ai/vitest.edge.config.js b/packages/ai/vitest.edge.config.js
index 32fbed1c6dae..ab1a7969f6a8 100644
--- a/packages/ai/vitest.edge.config.js
+++ b/packages/ai/vitest.edge.config.js
@@ -1,7 +1,15 @@
import { defineConfig } from 'vite';
+import { readFileSync } from 'node:fs';
+
+const version = JSON.parse(
+ readFileSync(new URL('./package.json', import.meta.url), 'utf-8'),
+).version;
// https://vitejs.dev/config/
export default defineConfig({
+ define: {
+ __PACKAGE_VERSION__: JSON.stringify(version),
+ },
test: {
environment: 'edge-runtime',
include: ['**/*.test.ts{,x}'],
diff --git a/packages/ai/vitest.node.config.js b/packages/ai/vitest.node.config.js
index 2518577b8137..091e0fbe8374 100644
--- a/packages/ai/vitest.node.config.js
+++ b/packages/ai/vitest.node.config.js
@@ -1,7 +1,15 @@
import { defineConfig } from 'vite';
+import { readFileSync } from 'node:fs';
+
+const version = JSON.parse(
+ readFileSync(new URL('./package.json', import.meta.url), 'utf-8'),
+).version;
// https://vitejs.dev/config/
export default defineConfig({
+ define: {
+ __PACKAGE_VERSION__: JSON.stringify(version),
+ },
test: {
environment: 'node',
include: ['**/*.test.ts{,x}'],
diff --git a/packages/amazon-bedrock/CHANGELOG.md b/packages/amazon-bedrock/CHANGELOG.md
index 8ebb51169aed..eb3a98465487 100644
--- a/packages/amazon-bedrock/CHANGELOG.md
+++ b/packages/amazon-bedrock/CHANGELOG.md
@@ -1,5 +1,73 @@
# @ai-sdk/amazon-bedrock
+## 3.1.0-beta.3
+
+### Patch Changes
+
+- ed329cb: feat: `Provider-V3`
+- 522f6b8: feat: `ImageModelV3`
+- Updated dependencies [ed329cb]
+- Updated dependencies [522f6b8]
+ - @ai-sdk/anthropic@2.1.0-beta.3
+ - @ai-sdk/provider@2.1.0-beta.2
+ - @ai-sdk/provider-utils@3.1.0-beta.3
+
+## 3.1.0-beta.2
+
+### Patch Changes
+
+- 0c4822d: feat: `EmbeddingModelV3`
+- 1cad0ab: feat: add provider version to user-agent header
+- Updated dependencies [0c4822d]
+- Updated dependencies [1cad0ab]
+ - @ai-sdk/provider@2.1.0-beta.1
+ - @ai-sdk/anthropic@2.1.0-beta.2
+ - @ai-sdk/provider-utils@3.1.0-beta.2
+
+## 3.1.0-beta.1
+
+### Patch Changes
+
+- Updated dependencies [953d0f2]
+- Updated dependencies [cbb1d35]
+ - @ai-sdk/test-server@1.0.0-beta.0
+ - @ai-sdk/provider-utils@3.1.0-beta.1
+ - @ai-sdk/anthropic@2.1.0-beta.1
+
+## 3.1.0-beta.0
+
+### Minor Changes
+
+- 78928cb: release: start 5.1 beta
+
+### Patch Changes
+
+- Updated dependencies [78928cb]
+ - @ai-sdk/anthropic@2.1.0-beta.0
+ - @ai-sdk/provider@2.1.0-beta.0
+ - @ai-sdk/provider-utils@3.1.0-beta.0
+
+## 3.0.22
+
+### Patch Changes
+
+- 32f3cef: feat: adding user-agent to all packages that use global fetch directly
+
+## 3.0.21
+
+### Patch Changes
+
+- Updated dependencies [da92132]
+ - @ai-sdk/anthropic@2.0.17
+
+## 3.0.20
+
+### Patch Changes
+
+- Updated dependencies [0294b58]
+ - @ai-sdk/provider-utils@3.0.9
+ - @ai-sdk/anthropic@2.0.16
+
## 3.0.19
### Patch Changes
diff --git a/packages/amazon-bedrock/package.json b/packages/amazon-bedrock/package.json
index 16f6603a7546..965f9321b34e 100644
--- a/packages/amazon-bedrock/package.json
+++ b/packages/amazon-bedrock/package.json
@@ -1,6 +1,6 @@
{
"name": "@ai-sdk/amazon-bedrock",
- "version": "3.0.19",
+ "version": "3.1.0-beta.3",
"license": "Apache-2.0",
"sideEffects": false,
"main": "./dist/index.js",
@@ -35,6 +35,7 @@
"@ai-sdk/anthropic": "workspace:*",
"@ai-sdk/provider": "workspace:*",
"@ai-sdk/provider-utils": "workspace:*",
+ "@ai-sdk/test-server": "workspace:*",
"@smithy/eventstream-codec": "^4.0.1",
"@smithy/util-utf8": "^4.0.0",
"aws4fetch": "^1.0.20"
diff --git a/packages/amazon-bedrock/src/bedrock-chat-language-model.test.ts b/packages/amazon-bedrock/src/bedrock-chat-language-model.test.ts
index 8379ec0fa1f0..f5bcbe933e58 100644
--- a/packages/amazon-bedrock/src/bedrock-chat-language-model.test.ts
+++ b/packages/amazon-bedrock/src/bedrock-chat-language-model.test.ts
@@ -1,8 +1,6 @@
import { LanguageModelV2Prompt } from '@ai-sdk/provider';
-import {
- createTestServer,
- convertReadableStreamToArray,
-} from '@ai-sdk/provider-utils/test';
+import { createTestServer } from '@ai-sdk/test-server/with-vitest';
+import { convertReadableStreamToArray } from '@ai-sdk/provider-utils/test';
import { BedrockChatLanguageModel } from './bedrock-chat-language-model';
import { beforeEach, describe, expect, vi, it } from 'vitest';
import { injectFetchHeaders } from './inject-fetch-headers';
diff --git a/packages/amazon-bedrock/src/bedrock-embedding-model.test.ts b/packages/amazon-bedrock/src/bedrock-embedding-model.test.ts
index bff436d1533a..0fb10b6dc3ec 100644
--- a/packages/amazon-bedrock/src/bedrock-embedding-model.test.ts
+++ b/packages/amazon-bedrock/src/bedrock-embedding-model.test.ts
@@ -1,4 +1,4 @@
-import { createTestServer } from '@ai-sdk/provider-utils/test';
+import { createTestServer } from '@ai-sdk/test-server/with-vitest';
import { BedrockEmbeddingModel } from './bedrock-embedding-model';
import { injectFetchHeaders } from './inject-fetch-headers';
import { beforeEach, describe, expect, it } from 'vitest';
diff --git a/packages/amazon-bedrock/src/bedrock-embedding-model.ts b/packages/amazon-bedrock/src/bedrock-embedding-model.ts
index 66953ccb4760..bdcb2453689d 100644
--- a/packages/amazon-bedrock/src/bedrock-embedding-model.ts
+++ b/packages/amazon-bedrock/src/bedrock-embedding-model.ts
@@ -1,5 +1,5 @@
import {
- EmbeddingModelV2,
+ EmbeddingModelV3,
TooManyEmbeddingValuesForCallError,
} from '@ai-sdk/provider';
import {
@@ -25,10 +25,10 @@ type BedrockEmbeddingConfig = {
fetch?: FetchFunction;
};
-type DoEmbedResponse = Awaited['doEmbed']>>;
+type DoEmbedResponse = Awaited['doEmbed']>>;
-export class BedrockEmbeddingModel implements EmbeddingModelV2 {
- readonly specificationVersion = 'v2';
+export class BedrockEmbeddingModel implements EmbeddingModelV3 {
+ readonly specificationVersion = 'v3';
readonly provider = 'amazon-bedrock';
readonly maxEmbeddingsPerCall = 1;
readonly supportsParallelCalls = true;
@@ -49,7 +49,7 @@ export class BedrockEmbeddingModel implements EmbeddingModelV2 {
abortSignal,
providerOptions,
}: Parameters<
- EmbeddingModelV2['doEmbed']
+ EmbeddingModelV3['doEmbed']
>[0]): Promise {
if (values.length > this.maxEmbeddingsPerCall) {
throw new TooManyEmbeddingValuesForCallError({
diff --git a/packages/amazon-bedrock/src/bedrock-image-model.test.ts b/packages/amazon-bedrock/src/bedrock-image-model.test.ts
index ac935b38ff51..8dcbcfc14b5b 100644
--- a/packages/amazon-bedrock/src/bedrock-image-model.test.ts
+++ b/packages/amazon-bedrock/src/bedrock-image-model.test.ts
@@ -1,4 +1,4 @@
-import { createTestServer } from '@ai-sdk/provider-utils/test';
+import { createTestServer } from '@ai-sdk/test-server/with-vitest';
import { createAmazonBedrock } from './bedrock-provider';
import { BedrockImageModel } from './bedrock-image-model';
import { injectFetchHeaders } from './inject-fetch-headers';
diff --git a/packages/amazon-bedrock/src/bedrock-image-model.ts b/packages/amazon-bedrock/src/bedrock-image-model.ts
index 966580a763d1..cd5a08ba9e14 100644
--- a/packages/amazon-bedrock/src/bedrock-image-model.ts
+++ b/packages/amazon-bedrock/src/bedrock-image-model.ts
@@ -1,4 +1,4 @@
-import { ImageModelV2, ImageModelV2CallWarning } from '@ai-sdk/provider';
+import { ImageModelV3, ImageModelV3CallWarning } from '@ai-sdk/provider';
import {
FetchFunction,
Resolvable,
@@ -24,8 +24,8 @@ type BedrockImageModelConfig = {
};
};
-export class BedrockImageModel implements ImageModelV2 {
- readonly specificationVersion = 'v2';
+export class BedrockImageModel implements ImageModelV3 {
+ readonly specificationVersion = 'v3';
readonly provider = 'amazon-bedrock';
get maxImagesPerCall(): number {
@@ -51,10 +51,10 @@ export class BedrockImageModel implements ImageModelV2 {
providerOptions,
headers,
abortSignal,
- }: Parameters[0]): Promise<
- Awaited>
+ }: Parameters[0]): Promise<
+ Awaited>
> {
- const warnings: Array = [];
+ const warnings: Array = [];
const [width, height] = size ? size.split('x').map(Number) : [];
const args = {
taskType: 'TEXT_IMAGE',
diff --git a/packages/amazon-bedrock/src/bedrock-provider.test.ts b/packages/amazon-bedrock/src/bedrock-provider.test.ts
index 978d121e8946..1bdb3298e8d3 100644
--- a/packages/amazon-bedrock/src/bedrock-provider.test.ts
+++ b/packages/amazon-bedrock/src/bedrock-provider.test.ts
@@ -60,6 +60,10 @@ vi.mock('@ai-sdk/provider-utils', async importOriginal => {
};
});
+vi.mock('./version', () => ({
+ VERSION: '0.0.0-test',
+}));
+
// Import mocked modules to get references
import {
createSigV4FetchFunction,
@@ -89,14 +93,17 @@ describe('AmazonBedrockProvider', () => {
const constructorCall = BedrockChatLanguageModelMock.mock.calls[0];
expect(constructorCall[0]).toBe('anthropic.claude-v2');
- expect(constructorCall[1].headers).toEqual({});
+ expect(constructorCall[1].headers()).toMatchObject({});
+ expect(constructorCall[1].headers()['user-agent']).toContain(
+ 'ai-sdk/amazon-bedrock/0.0.0-test',
+ );
expect(constructorCall[1].baseUrl()).toBe(
'https://bedrock-runtime.us-east-1.amazonaws.com',
);
});
it('should create a provider instance with custom options', () => {
- const customHeaders = { 'Custom-Header': 'value' };
+ const customHeaders = { 'custom-header': 'value' };
const options = {
region: 'eu-west-1',
baseURL: 'https://custom.url',
@@ -107,7 +114,10 @@ describe('AmazonBedrockProvider', () => {
provider('anthropic.claude-v2');
const constructorCall = BedrockChatLanguageModelMock.mock.calls[0];
- expect(constructorCall[1].headers).toEqual(customHeaders);
+ expect(constructorCall[1].headers()).toMatchObject(customHeaders);
+ expect(constructorCall[1].headers()['user-agent']).toContain(
+ 'ai-sdk/amazon-bedrock/0.0.0-test',
+ );
expect(constructorCall[1].baseUrl()).toBe('https://custom.url');
});
@@ -126,7 +136,10 @@ describe('AmazonBedrockProvider', () => {
const constructorCall = BedrockChatLanguageModelMock.mock.calls[0];
expect(constructorCall[0]).toBe('anthropic.claude-v2');
- expect(constructorCall[1].headers).toEqual({});
+ expect(constructorCall[1].headers()).toMatchObject({});
+ expect(constructorCall[1].headers()['user-agent']).toContain(
+ 'ai-sdk/amazon-bedrock/0.0.0-test',
+ );
expect(constructorCall[1].baseUrl()).toBe(
'https://bedrock-runtime.us-east-1.amazonaws.com',
);
@@ -152,7 +165,7 @@ describe('AmazonBedrockProvider', () => {
});
it('should pass headers to embedding model', () => {
- const customHeaders = { 'Custom-Header': 'value' };
+ const customHeaders = { 'custom-header': 'value' };
const provider = createAmazonBedrock({
headers: customHeaders,
});
@@ -160,7 +173,10 @@ describe('AmazonBedrockProvider', () => {
provider.embedding('amazon.titan-embed-text-v1');
const constructorCall = BedrockEmbeddingModelMock.mock.calls[0];
- expect(constructorCall[1].headers).toEqual(customHeaders);
+ expect(constructorCall[1].headers()).toMatchObject(customHeaders);
+ expect(constructorCall[1].headers()['user-agent']).toContain(
+ 'ai-sdk/amazon-bedrock/0.0.0-test',
+ );
});
it('should throw error when called with new keyword', () => {
@@ -190,7 +206,10 @@ describe('AmazonBedrockProvider', () => {
const constructorCall = BedrockChatLanguageModelMock.mock.calls[0];
expect(constructorCall[0]).toBe('anthropic.claude-v2');
- expect(constructorCall[1].headers).toEqual({});
+ expect(constructorCall[1].headers()).toMatchObject({});
+ expect(constructorCall[1].headers()['user-agent']).toContain(
+ 'ai-sdk/amazon-bedrock/',
+ );
expect(constructorCall[1].baseUrl()).toBe(
'https://bedrock-runtime.us-east-1.amazonaws.com',
);
@@ -306,16 +325,19 @@ describe('AmazonBedrockProvider', () => {
const provider = createAmazonBedrock({
apiKey: 'test-api-key',
region: 'us-east-1',
- headers: { 'Custom-Header': 'value' },
+ headers: { 'custom-header': 'value' },
});
provider.embedding('amazon.titan-embed-text-v1');
const constructorCall = BedrockEmbeddingModelMock.mock.calls[0];
expect(constructorCall[0]).toBe('amazon.titan-embed-text-v1');
- expect(constructorCall[1].headers).toEqual({
- 'Custom-Header': 'value',
+ expect(constructorCall[1].headers()).toMatchObject({
+ 'custom-header': 'value',
});
+ expect(constructorCall[1].headers()['user-agent']).toContain(
+ 'ai-sdk/amazon-bedrock/0.0.0-test',
+ );
expect(mockCreateApiKeyFetchFunction).toHaveBeenCalledWith(
'test-api-key',
undefined,
@@ -326,16 +348,19 @@ describe('AmazonBedrockProvider', () => {
const provider = createAmazonBedrock({
apiKey: 'test-api-key',
region: 'us-east-1',
- headers: { 'Custom-Header': 'value' },
+ headers: { 'custom-header': 'value' },
});
provider.image('amazon.titan-image-generator');
const constructorCall = BedrockImageModelMock.mock.calls[0];
expect(constructorCall[0]).toBe('amazon.titan-image-generator');
- expect(constructorCall[1].headers).toEqual({
- 'Custom-Header': 'value',
+ expect(constructorCall[1].headers()).toMatchObject({
+ 'custom-header': 'value',
});
+ expect(constructorCall[1].headers()['user-agent']).toContain(
+ 'ai-sdk/amazon-bedrock/0.0.0-test',
+ );
expect(mockCreateApiKeyFetchFunction).toHaveBeenCalledWith(
'test-api-key',
undefined,
diff --git a/packages/amazon-bedrock/src/bedrock-provider.ts b/packages/amazon-bedrock/src/bedrock-provider.ts
index 5c58f8a5e732..7fca538d04bb 100644
--- a/packages/amazon-bedrock/src/bedrock-provider.ts
+++ b/packages/amazon-bedrock/src/bedrock-provider.ts
@@ -1,8 +1,8 @@
import {
- EmbeddingModelV2,
- ImageModelV2,
+ EmbeddingModelV3,
+ ImageModelV3,
LanguageModelV2,
- ProviderV2,
+ ProviderV3,
} from '@ai-sdk/provider';
import {
FetchFunction,
@@ -10,7 +10,9 @@ import {
loadOptionalSetting,
loadSetting,
withoutTrailingSlash,
+ withUserAgentSuffix,
} from '@ai-sdk/provider-utils';
+import { VERSION } from './version';
import { anthropicTools } from '@ai-sdk/anthropic/internal';
import { BedrockChatLanguageModel } from './bedrock-chat-language-model';
import { BedrockChatModelId } from './bedrock-chat-options';
@@ -102,22 +104,22 @@ and `sessionToken` settings.
generateId?: () => string;
}
-export interface AmazonBedrockProvider extends ProviderV2 {
+export interface AmazonBedrockProvider extends ProviderV3 {
(modelId: BedrockChatModelId): LanguageModelV2;
languageModel(modelId: BedrockChatModelId): LanguageModelV2;
- embedding(modelId: BedrockEmbeddingModelId): EmbeddingModelV2
- OpenAI File Search Block-Based Streaming Test -
+
+ sendMessage({ text })} />
);
diff --git a/examples/next-openai/app/test-openai-image-generation/page.tsx b/examples/next-openai/app/test-openai-image-generation/page.tsx
new file mode 100644
index 000000000000..76f5f707551e
--- /dev/null
+++ b/examples/next-openai/app/test-openai-image-generation/page.tsx
@@ -0,0 +1,38 @@
+'use client';
+
+import { OpenAIImageGenerationMessage } from '@/app/api/chat-openai-image-generation/route';
+import ChatInput from '@/component/chat-input';
+import ImageGenerationView from '@/component/openai-image-generation-view';
+import { useChat } from '@ai-sdk/react';
+import { DefaultChatTransport } from 'ai';
+
+export default function TestOpenAIWebSearch() {
+ const { status, sendMessage, messages } =
+ useChatOpenAI File Search Test
{messages.map(message => (
{message.role === 'user' ? 'User: ' : 'AI: '}
{message.parts.map((part, index) => {
- if (part.type === 'text') {
- return
))}
- {(status === 'submitted' || status === 'streaming') && (
- {part.text}
;
- }
-
- if (part.type === 'tool-file_search') {
- if (part.state === 'input-available') {
- return (
-
- {JSON.stringify(part.input, null, 2)}
-
- );
- }
- if (part.state === 'output-available') {
- return (
-
- {JSON.stringify(part.input, null, 2)}
- {`\n\nDONE - File search completed`}
-
- );
- }
+ switch (part.type) {
+ case 'text':
+ return {part.text}
;
+ case 'tool-file_search':
+ return
- {status === 'submitted' &&
- )}
-
- {error && (
- Loading...
}
-
-
-
- )}
-
An error occurred.
-
-
+ sendMessage({ text })} />
+
+ );
+}
diff --git a/examples/next-openai/component/openai-code-interpreter-view.tsx b/examples/next-openai/component/openai-code-interpreter-view.tsx
new file mode 100644
index 000000000000..6289f5e3a639
--- /dev/null
+++ b/examples/next-openai/component/openai-code-interpreter-view.tsx
@@ -0,0 +1,52 @@
+import { openai } from '@ai-sdk/openai';
+import { UIToolInvocation } from 'ai';
+
+export default function CodeInterpreterView({
+ invocation,
+}: {
+ invocation: UIToolInvocationOpenAI Image Generation Test
+ + {messages.map(message => ( +
+ {message.role === 'user' ? 'User: ' : 'AI: '}
+ {message.parts.map((part, index) => {
+ switch (part.type) {
+ case 'text':
+ return
+ ))}
+
+ {part.text}
;
+ case 'tool-image_generation':
+ return
+
+ );
+ }
+}
diff --git a/examples/next-openai/component/openai-file-search-view.tsx b/examples/next-openai/component/openai-file-search-view.tsx
new file mode 100644
index 000000000000..b27e34c5d0f9
--- /dev/null
+++ b/examples/next-openai/component/openai-file-search-view.tsx
@@ -0,0 +1,42 @@
+import { openai } from '@ai-sdk/openai';
+import { UIToolInvocation } from 'ai';
+
+export default function FileSearchView({
+ invocation,
+}: {
+ invocation: UIToolInvocation
+
+
+
+ {invocation.input.containerId}
+
+
+
+
+
+
+
+ Code:
+
+
+ {invocation.input.code}
+
+
+
+
+ Output:
+
+
+ {invocation.output.outputs?.map((output, index) => (
+
+
+ {output.type === 'logs' && (
+ 
}
+
+ ))}
+
+
+ {output.logs}
+
+
+ )}
+ {output.type === 'image' &&
+ Searching...
+
+ );
+ case 'output-available':
+ return (
+
+
+ );
+ }
+}
diff --git a/examples/next-openai/component/openai-image-generation-view.tsx b/examples/next-openai/component/openai-image-generation-view.tsx
new file mode 100644
index 000000000000..f4c4b1e8ced4
--- /dev/null
+++ b/examples/next-openai/component/openai-image-generation-view.tsx
@@ -0,0 +1,23 @@
+import { openai } from '@ai-sdk/openai';
+import { UIToolInvocation } from 'ai';
+
+export default function ImageGenerationView({
+ invocation,
+}: {
+ invocation: UIToolInvocation
+ Queries:
+
+ -
+ {invocation.output.queries.map((query, index) => (
+
- - {query} + ))} +
+ Results:
+
+
+ {invocation.output.results?.map((result, index) => (
+
+
+
+ ))}
+
+ {JSON.stringify(result, null, 2)}
+
+
+ Generating image...
+
+ );
+ case 'output-available':
+ return (
+
+ 
+
+ );
+ }
+}
diff --git a/examples/next/app/api/chat/route.ts b/examples/next/app/api/chat/route.ts
index 54b5b11e46c6..1ec9ee37f75c 100644
--- a/examples/next/app/api/chat/route.ts
+++ b/examples/next/app/api/chat/route.ts
@@ -1,5 +1,4 @@
import type { MyUIMessage } from '@/util/chat-schema';
-import { openai } from '@ai-sdk/openai';
import { readChat, saveChat } from '@util/chat-store';
import { convertToModelMessages, generateId, streamText } from 'ai';
import { after } from 'next/server';
@@ -57,7 +56,7 @@ export async function POST(req: Request) {
saveChat({ id, messages, activeStreamId: null });
const result = streamText({
- model: openai('gpt-4o-mini'),
+ model: 'openai/gpt-5-mini',
messages: convertToModelMessages(messages),
abortSignal: req.signal,
});
diff --git a/examples/next/package.json b/examples/next/package.json
index b8d07801f9c7..1a336b3a68d7 100644
--- a/examples/next/package.json
+++ b/examples/next/package.json
@@ -9,7 +9,6 @@
"lint": "next lint"
},
"dependencies": {
- "@ai-sdk/openai": "workspace:*",
"@ai-sdk/react": "workspace:*",
"@vercel/blob": "^0.26.0",
"ai": "workspace:*",
diff --git a/examples/next/util/chat-schema.ts b/examples/next/util/chat-schema.ts
index 8c49a52f5fce..ff6e30930578 100644
--- a/examples/next/util/chat-schema.ts
+++ b/examples/next/util/chat-schema.ts
@@ -1,5 +1,5 @@
import { UIDataTypes, UIMessage } from 'ai';
-import { z } from 'zod/v4';
+import { z } from 'zod';
export const myMessageMetadataSchema = z.object({
createdAt: z.number(),
diff --git a/examples/nuxt-openai/pages/use-object/index.vue b/examples/nuxt-openai/pages/use-object/index.vue
new file mode 100644
index 000000000000..737b01067ee2
--- /dev/null
+++ b/examples/nuxt-openai/pages/use-object/index.vue
@@ -0,0 +1,69 @@
+
+
+
+
+
+
+
+
diff --git a/examples/nuxt-openai/server/api/use-chat-tools.ts b/examples/nuxt-openai/server/api/use-chat-tools.ts
index 71c4d1f3dfb0..d4d0c32fc436 100644
--- a/examples/nuxt-openai/server/api/use-chat-tools.ts
+++ b/examples/nuxt-openai/server/api/use-chat-tools.ts
@@ -1,6 +1,6 @@
import { createOpenAI } from '@ai-sdk/openai';
import { convertToModelMessages, stepCountIs, streamText } from 'ai';
-import { z } from 'zod/v4';
+import { z } from 'zod';
export default defineLazyEventHandler(async () => {
const openai = createOpenAI({
diff --git a/examples/nuxt-openai/server/api/use-object.ts b/examples/nuxt-openai/server/api/use-object.ts
new file mode 100644
index 000000000000..4e7a59deb3e3
--- /dev/null
+++ b/examples/nuxt-openai/server/api/use-object.ts
@@ -0,0 +1,22 @@
+import { createOpenAI } from '@ai-sdk/openai';
+import { streamObject } from 'ai';
+import { notificationSchema } from '~/shared/notification-schema';
+
+export default defineLazyEventHandler(async () => {
+ const apiKey = useRuntimeConfig().openaiApiKey;
+ if (!apiKey) throw new Error('Missing OpenAI API key');
+ const openai = createOpenAI({ apiKey });
+
+ return defineEventHandler(async (event: any) => {
+ const context = await readBody(event);
+
+ // Stream generated notifications as objects
+ const result = streamObject({
+ model: openai('gpt-4.1'),
+ prompt: `Generate 5 notifications for a messages app in this context: ${context}`,
+ schema: notificationSchema,
+ });
+
+ return result.toTextStreamResponse();
+ });
+});
diff --git a/examples/nuxt-openai/shared/notification-schema.ts b/examples/nuxt-openai/shared/notification-schema.ts
new file mode 100644
index 000000000000..932f6eb3de7f
--- /dev/null
+++ b/examples/nuxt-openai/shared/notification-schema.ts
@@ -0,0 +1,11 @@
+import { z } from 'zod';
+
+export const notificationSchema = z.object({
+ notifications: z.array(
+ z.object({
+ name: z.string().describe('Name of a fictional person.'),
+ message: z.string().describe('Message. Do not use emojis or links.'),
+ minutesAgo: z.number(),
+ }),
+ ),
+});
diff --git a/examples/sveltekit-openai/src/routes/api/chat/+server.ts b/examples/sveltekit-openai/src/routes/api/chat/+server.ts
index c191f3a45716..1446b9e0fe5a 100644
--- a/examples/sveltekit-openai/src/routes/api/chat/+server.ts
+++ b/examples/sveltekit-openai/src/routes/api/chat/+server.ts
@@ -1,7 +1,7 @@
import { env } from '$env/dynamic/private';
import { createOpenAI } from '@ai-sdk/openai';
import { convertToModelMessages, streamText, stepCountIs } from 'ai';
-import { z } from 'zod/v4';
+import { z } from 'zod';
const openai = createOpenAI({
apiKey: env?.OPENAI_API_KEY,
diff --git a/examples/sveltekit-openai/src/routes/structured-object/schema.ts b/examples/sveltekit-openai/src/routes/structured-object/schema.ts
index 539f51299117..72104f1e8b25 100644
--- a/examples/sveltekit-openai/src/routes/structured-object/schema.ts
+++ b/examples/sveltekit-openai/src/routes/structured-object/schema.ts
@@ -1,4 +1,4 @@
-import { z } from 'zod/v4';
+import { z } from 'zod';
// define a schema for the notifications
export const notificationSchema = z.object({
diff --git a/packages/ai/CHANGELOG.md b/packages/ai/CHANGELOG.md
index 7318ca092ec9..ca74e89ccdff 100644
--- a/packages/ai/CHANGELOG.md
+++ b/packages/ai/CHANGELOG.md
@@ -1,5 +1,138 @@
# ai
+## 5.1.0-beta.9
+
+### Patch Changes
+
+- ed329cb: feat: `Provider-V3`
+- 177b475: fix(ai): download files when intermediate file cannot be downloaded
+- 522f6b8: feat: `ImageModelV3`
+- Updated dependencies [aaf5ebf]
+- Updated dependencies [ed329cb]
+- Updated dependencies [522f6b8]
+ - @ai-sdk/gateway@1.1.0-beta.6
+ - @ai-sdk/provider@2.1.0-beta.2
+ - @ai-sdk/provider-utils@3.1.0-beta.3
+
+## 5.1.0-beta.8
+
+### Patch Changes
+
+- 7eca093: fix(ai): update `uiMessageChunkSchema` to satisfy the `UIMessageChunk` type
+
+## 5.1.0-beta.7
+
+### Patch Changes
+
+- 5a4e732: Export `parseJsonEventStream` and `uiMessageChunkSchema` from "ai" package
+
+## 5.1.0-beta.6
+
+### Patch Changes
+
+- 0c4822d: feat: `EmbeddingModelV3`
+- Updated dependencies [5d21222]
+- Updated dependencies [0c4822d]
+- Updated dependencies [1cad0ab]
+ - @ai-sdk/gateway@1.1.0-beta.5
+ - @ai-sdk/provider@2.1.0-beta.1
+ - @ai-sdk/provider-utils@3.1.0-beta.2
+
+## 5.1.0-beta.5
+
+### Patch Changes
+
+- Updated dependencies [7ccb36f]
+- Updated dependencies [6c766ef]
+ - @ai-sdk/gateway@1.1.0-beta.4
+
+## 5.1.0-beta.4
+
+### Patch Changes
+
+- Updated dependencies [ea9ca31]
+ - @ai-sdk/gateway@1.1.0-beta.3
+
+## 5.1.0-beta.3
+
+### Patch Changes
+
+- Updated dependencies [3e83633]
+- Updated dependencies [f83903d]
+ - @ai-sdk/gateway@1.1.0-beta.2
+
+## 5.1.0-beta.2
+
+### Patch Changes
+
+- Updated dependencies [953d0f2]
+- Updated dependencies [cbb1d35]
+ - @ai-sdk/test-server@1.0.0-beta.0
+ - @ai-sdk/provider-utils@3.1.0-beta.1
+ - @ai-sdk/gateway@1.1.0-beta.1
+
+## 5.1.0-beta.1
+
+### Patch Changes
+
+- a7f6f81: Add safeValidateUIMessages utility to validate UI messages without throwing, returning a success/failure result object like Zod’s safeParse
+
+## 5.1.0-beta.0
+
+### Minor Changes
+
+- 78928cb: release: start 5.1 beta
+
+### Patch Changes
+
+- Updated dependencies [78928cb]
+ - @ai-sdk/gateway@1.1.0-beta.0
+ - @ai-sdk/provider@2.1.0-beta.0
+ - @ai-sdk/provider-utils@3.1.0-beta.0
+
+## 5.0.45
+
+### Patch Changes
+
+- 76024fc: fix(ai): fix static tool call and result detection when dynamic is undefined
+- 93d8b60: fix(ai): do not filter zero-length text parts that have provider options
+- d8eb31f: fix(ai): fix webp image detection from base64
+
+## 5.0.44
+
+### Patch Changes
+
+- Updated dependencies [f49f924]
+ - @ai-sdk/gateway@1.0.23
+
+## 5.0.43
+
+### Patch Changes
+
+- 0294b58: feat(ai): set `ai`, `@ai-sdk/provider-utils`, and runtime in `user-agent` header
+- Updated dependencies [0294b58]
+ - @ai-sdk/provider-utils@3.0.9
+ - @ai-sdk/gateway@1.0.22
+
+## 5.0.42
+
+### Patch Changes
+
+- de5c066: fix(ai): forwarded providerExecuted flag in validateUIMessages
+
+## 5.0.41
+
+### Patch Changes
+
+- cd91e4b: fix(ai): use correct type for reasoning outputs
+
+## 5.0.40
+
+### Patch Changes
+
+- Updated dependencies [4ee3719]
+ - @ai-sdk/gateway@1.0.21
+
## 5.0.39
### Patch Changes
diff --git a/packages/ai/package.json b/packages/ai/package.json
index 1857be2eb7ce..760a58fa96a7 100644
--- a/packages/ai/package.json
+++ b/packages/ai/package.json
@@ -1,6 +1,6 @@
{
"name": "ai",
- "version": "5.0.39",
+ "version": "5.1.0-beta.9",
"description": "AI SDK by Vercel - The AI Toolkit for TypeScript and JavaScript",
"license": "Apache-2.0",
"sideEffects": false,
@@ -54,6 +54,7 @@
}
},
"dependencies": {
+ "@ai-sdk/test-server": "workspace:*",
"@ai-sdk/gateway": "workspace:*",
"@ai-sdk/provider": "workspace:*",
"@ai-sdk/provider-utils": "workspace:*",
diff --git a/packages/ai/src/embed/__snapshots__/embed-many.test.ts.snap b/packages/ai/src/embed/__snapshots__/embed-many.test.ts.snap
index 449bc194042b..e028b64865ec 100644
--- a/packages/ai/src/embed/__snapshots__/embed-many.test.ts.snap
+++ b/packages/ai/src/embed/__snapshots__/embed-many.test.ts.snap
@@ -27,6 +27,7 @@ exports[`telemetry > should not record telemetry inputs / outputs when disabled
"ai.model.id": "mock-model-id",
"ai.model.provider": "mock-provider",
"ai.operationId": "ai.embedMany",
+ "ai.request.headers.user-agent": "ai/0.0.0-test",
"ai.settings.maxRetries": 2,
"ai.usage.tokens": 10,
"operation.name": "ai.embedMany",
@@ -39,6 +40,7 @@ exports[`telemetry > should not record telemetry inputs / outputs when disabled
"ai.model.id": "mock-model-id",
"ai.model.provider": "mock-provider",
"ai.operationId": "ai.embedMany.doEmbed",
+ "ai.request.headers.user-agent": "ai/0.0.0-test",
"ai.settings.maxRetries": 2,
"ai.usage.tokens": 10,
"operation.name": "ai.embedMany.doEmbed",
@@ -61,6 +63,7 @@ exports[`telemetry > should record telemetry data when enabled (multiple calls p
"ai.model.id": "mock-model-id",
"ai.model.provider": "mock-provider",
"ai.operationId": "ai.embedMany",
+ "ai.request.headers.user-agent": "ai/0.0.0-test",
"ai.settings.maxRetries": 2,
"ai.telemetry.functionId": "test-function-id",
"ai.telemetry.metadata.test1": "value1",
@@ -86,6 +89,7 @@ exports[`telemetry > should record telemetry data when enabled (multiple calls p
"ai.model.id": "mock-model-id",
"ai.model.provider": "mock-provider",
"ai.operationId": "ai.embedMany.doEmbed",
+ "ai.request.headers.user-agent": "ai/0.0.0-test",
"ai.settings.maxRetries": 2,
"ai.telemetry.functionId": "test-function-id",
"ai.telemetry.metadata.test1": "value1",
@@ -109,6 +113,7 @@ exports[`telemetry > should record telemetry data when enabled (multiple calls p
"ai.model.id": "mock-model-id",
"ai.model.provider": "mock-provider",
"ai.operationId": "ai.embedMany.doEmbed",
+ "ai.request.headers.user-agent": "ai/0.0.0-test",
"ai.settings.maxRetries": 2,
"ai.telemetry.functionId": "test-function-id",
"ai.telemetry.metadata.test1": "value1",
@@ -138,6 +143,7 @@ exports[`telemetry > should record telemetry data when enabled (single call path
"ai.model.id": "mock-model-id",
"ai.model.provider": "mock-provider",
"ai.operationId": "ai.embedMany",
+ "ai.request.headers.user-agent": "ai/0.0.0-test",
"ai.settings.maxRetries": 2,
"ai.telemetry.functionId": "test-function-id",
"ai.telemetry.metadata.test1": "value1",
@@ -164,6 +170,7 @@ exports[`telemetry > should record telemetry data when enabled (single call path
"ai.model.id": "mock-model-id",
"ai.model.provider": "mock-provider",
"ai.operationId": "ai.embedMany.doEmbed",
+ "ai.request.headers.user-agent": "ai/0.0.0-test",
"ai.settings.maxRetries": 2,
"ai.telemetry.functionId": "test-function-id",
"ai.telemetry.metadata.test1": "value1",
diff --git a/packages/ai/src/embed/__snapshots__/embed.test.ts.snap b/packages/ai/src/embed/__snapshots__/embed.test.ts.snap
index cb41f8f95dba..eb32140c247c 100644
--- a/packages/ai/src/embed/__snapshots__/embed.test.ts.snap
+++ b/packages/ai/src/embed/__snapshots__/embed.test.ts.snap
@@ -9,6 +9,7 @@ exports[`telemetry > should not record telemetry inputs / outputs when disabled
"ai.model.id": "mock-model-id",
"ai.model.provider": "mock-provider",
"ai.operationId": "ai.embed",
+ "ai.request.headers.user-agent": "ai/0.0.0-test",
"ai.settings.maxRetries": 2,
"ai.usage.tokens": 10,
"operation.name": "ai.embed",
@@ -21,6 +22,7 @@ exports[`telemetry > should not record telemetry inputs / outputs when disabled
"ai.model.id": "mock-model-id",
"ai.model.provider": "mock-provider",
"ai.operationId": "ai.embed.doEmbed",
+ "ai.request.headers.user-agent": "ai/0.0.0-test",
"ai.settings.maxRetries": 2,
"ai.usage.tokens": 10,
"operation.name": "ai.embed.doEmbed",
@@ -39,6 +41,7 @@ exports[`telemetry > should record telemetry data when enabled 1`] = `
"ai.model.id": "mock-model-id",
"ai.model.provider": "mock-provider",
"ai.operationId": "ai.embed",
+ "ai.request.headers.user-agent": "ai/0.0.0-test",
"ai.settings.maxRetries": 2,
"ai.telemetry.functionId": "test-function-id",
"ai.telemetry.metadata.test1": "value1",
@@ -59,6 +62,7 @@ exports[`telemetry > should record telemetry data when enabled 1`] = `
"ai.model.id": "mock-model-id",
"ai.model.provider": "mock-provider",
"ai.operationId": "ai.embed.doEmbed",
+ "ai.request.headers.user-agent": "ai/0.0.0-test",
"ai.settings.maxRetries": 2,
"ai.telemetry.functionId": "test-function-id",
"ai.telemetry.metadata.test1": "value1",
diff --git a/packages/ai/src/embed/embed-many.test.ts b/packages/ai/src/embed/embed-many.test.ts
index 4acc73d736e4..05227a2f90c6 100644
--- a/packages/ai/src/embed/embed-many.test.ts
+++ b/packages/ai/src/embed/embed-many.test.ts
@@ -1,12 +1,18 @@
-import { EmbeddingModelV2 } from '@ai-sdk/provider';
+import { EmbeddingModelV3 } from '@ai-sdk/provider';
import assert from 'node:assert';
import { beforeEach, describe, expect, it, vi } from 'vitest';
-import { MockEmbeddingModelV2 } from '../test/mock-embedding-model-v2';
+import { MockEmbeddingModelV3 } from '../test/mock-embedding-model-v3';
import { MockTracer } from '../test/mock-tracer';
import { Embedding, EmbeddingModelUsage } from '../types';
import { createResolvablePromise } from '../util/create-resolvable-promise';
import { embedMany } from './embed-many';
+vi.mock('../version', () => {
+ return {
+ VERSION: '0.0.0-test',
+ };
+});
+
const dummyEmbeddings = [
[0.1, 0.2, 0.3],
[0.4, 0.5, 0.6],
@@ -31,7 +37,7 @@ describe('model.supportsParallelCalls', () => {
];
const embedManyPromise = embedMany({
- model: new MockEmbeddingModelV2({
+ model: new MockEmbeddingModelV3({
supportsParallelCalls: false,
maxEmbeddingsPerCall: 1,
doEmbed: async () => {
@@ -79,7 +85,7 @@ describe('model.supportsParallelCalls', () => {
];
const embedManyPromise = embedMany({
- model: new MockEmbeddingModelV2({
+ model: new MockEmbeddingModelV3({
supportsParallelCalls: true,
maxEmbeddingsPerCall: 1,
doEmbed: async () => {
@@ -128,7 +134,7 @@ describe('model.supportsParallelCalls', () => {
const embedManyPromise = embedMany({
maxParallelCalls: 2,
- model: new MockEmbeddingModelV2({
+ model: new MockEmbeddingModelV3({
supportsParallelCalls: true,
maxEmbeddingsPerCall: 1,
doEmbed: async () => {
@@ -169,7 +175,7 @@ describe('model.supportsParallelCalls', () => {
describe('result.embedding', () => {
it('should generate embeddings', async () => {
const result = await embedMany({
- model: new MockEmbeddingModelV2({
+ model: new MockEmbeddingModelV3({
maxEmbeddingsPerCall: 5,
doEmbed: mockEmbed(testValues, dummyEmbeddings),
}),
@@ -183,7 +189,7 @@ describe('result.embedding', () => {
let callCount = 0;
const result = await embedMany({
- model: new MockEmbeddingModelV2({
+ model: new MockEmbeddingModelV3({
maxEmbeddingsPerCall: 2,
doEmbed: async ({ values }) => {
switch (callCount++) {
@@ -209,7 +215,7 @@ describe('result.responses', () => {
it('should include responses in the result', async () => {
let callCount = 0;
const result = await embedMany({
- model: new MockEmbeddingModelV2({
+ model: new MockEmbeddingModelV3({
maxEmbeddingsPerCall: 1,
doEmbed: async ({ values }) => {
@@ -253,7 +259,7 @@ describe('result.responses', () => {
describe('result.values', () => {
it('should include values in the result', async () => {
const result = await embedMany({
- model: new MockEmbeddingModelV2({
+ model: new MockEmbeddingModelV3({
maxEmbeddingsPerCall: 5,
doEmbed: mockEmbed(testValues, dummyEmbeddings),
}),
@@ -269,7 +275,7 @@ describe('result.usage', () => {
let callCount = 0;
const result = await embedMany({
- model: new MockEmbeddingModelV2({
+ model: new MockEmbeddingModelV3({
maxEmbeddingsPerCall: 2,
doEmbed: async () => {
switch (callCount++) {
@@ -298,18 +304,21 @@ describe('result.usage', () => {
describe('options.headers', () => {
it('should set headers', async () => {
const result = await embedMany({
- model: new MockEmbeddingModelV2({
+ model: new MockEmbeddingModelV3({
maxEmbeddingsPerCall: 5,
doEmbed: async ({ headers }) => {
assert.deepStrictEqual(headers, {
'custom-request-header': 'request-header-value',
+ 'user-agent': 'ai/0.0.0-test',
});
return { embeddings: dummyEmbeddings };
},
}),
values: testValues,
- headers: { 'custom-request-header': 'request-header-value' },
+ headers: {
+ 'custom-request-header': 'request-header-value',
+ },
});
assert.deepStrictEqual(result.embeddings, dummyEmbeddings);
@@ -318,7 +327,7 @@ describe('options.headers', () => {
describe('options.providerOptions', () => {
it('should pass provider options to model', async () => {
- const model = new MockEmbeddingModelV2({
+ const model = new MockEmbeddingModelV3({
doEmbed: async ({ providerOptions }) => {
return { embeddings: [[1, 2, 3]] };
},
@@ -336,7 +345,9 @@ describe('options.providerOptions', () => {
expect(model.doEmbed).toHaveBeenCalledWith({
abortSignal: undefined,
- headers: undefined,
+ headers: {
+ 'user-agent': 'ai/0.0.0-test',
+ },
providerOptions: {
aProvider: { someKey: 'someValue' },
},
@@ -354,7 +365,7 @@ describe('telemetry', () => {
it('should not record any telemetry data when not explicitly enabled', async () => {
await embedMany({
- model: new MockEmbeddingModelV2({
+ model: new MockEmbeddingModelV3({
maxEmbeddingsPerCall: 5,
doEmbed: mockEmbed(testValues, dummyEmbeddings),
}),
@@ -368,7 +379,7 @@ describe('telemetry', () => {
let callCount = 0;
await embedMany({
- model: new MockEmbeddingModelV2({
+ model: new MockEmbeddingModelV3({
maxEmbeddingsPerCall: 2,
doEmbed: async ({ values }) => {
switch (callCount++) {
@@ -406,7 +417,7 @@ describe('telemetry', () => {
it('should record telemetry data when enabled (single call path)', async () => {
await embedMany({
- model: new MockEmbeddingModelV2({
+ model: new MockEmbeddingModelV3({
maxEmbeddingsPerCall: null,
doEmbed: mockEmbed(testValues, dummyEmbeddings, { tokens: 10 }),
}),
@@ -427,7 +438,7 @@ describe('telemetry', () => {
it('should not record telemetry inputs / outputs when disabled', async () => {
await embedMany({
- model: new MockEmbeddingModelV2({
+ model: new MockEmbeddingModelV3({
maxEmbeddingsPerCall: null,
doEmbed: mockEmbed(testValues, dummyEmbeddings, { tokens: 10 }),
}),
@@ -451,7 +462,7 @@ describe('result.providerMetadata', () => {
};
const result = await embedMany({
- model: new MockEmbeddingModelV2({
+ model: new MockEmbeddingModelV3({
supportsParallelCalls: false,
maxEmbeddingsPerCall: 3,
doEmbed: mockEmbed(
@@ -477,12 +488,12 @@ function mockEmbed
+ An error occurred. {{ error.message }}
+
+
+
+
+
+ Loading...
+
+
+
+
+
+
+
+
+
+
+
+
+
+ + {{ notification?.name }} +
++ {{ notification?.minutesAgo }} + minutes ago +
++ {{ notification?.message }} +
+