diff --git a/cog_safe_push/main.py b/cog_safe_push/main.py
index 6a31ff8..a892290 100644
--- a/cog_safe_push/main.py
+++ b/cog_safe_push/main.py
@@ -451,14 +451,14 @@ async def run_tasks(tasks: list[Task], parallel: int) -> None:
     log.info(f"Running tasks with parallelism {parallel}")
 
     semaphore = asyncio.Semaphore(parallel)
-    errors: list[Exception] = []
+    errors: list[tuple[Task, Exception]] = []
 
     async def run_with_semaphore(task: Task) -> None:
         async with semaphore:
             try:
                 await task.run()
             except Exception as e:
-                errors.append(e)
+                errors.append((task, e))
 
     # Create task coroutines and run them concurrently
     task_coroutines = [run_with_semaphore(task) for task in tasks]
@@ -467,11 +467,22 @@ async def run_with_semaphore(task: Task) -> None:
     await asyncio.gather(*task_coroutines, return_exceptions=True)
 
     if errors:
-        # If there are multiple errors, we'll raise the first one
-        # but log all of them
-        for error in errors[1:]:
-            log.error(f"Additional error occurred: {error}")
-        raise errors[0]
+        # Log all failures with their test case number and prediction URL
+        for task, error in errors:
+            prediction_index = getattr(task, "prediction_index", None)
+            prediction_url = getattr(task, "prediction_url", None)
+
+            prefix = f"[{prediction_index}] " if prediction_index is not None else ""
+            message = str(error)
+
+            if prediction_url:
+                log.error(
+                    f"{prefix}Test case failed: {message}; Prediction URL: {prediction_url}"
+                )
+            else:
+                log.error(f"{prefix}Test case failed: {message}")
+
+        raise errors[0][1]
 
 
 def parse_inputs(inputs_list: list[str]) -> dict[str, Any]:
diff --git a/cog_safe_push/predict.py b/cog_safe_push/predict.py
index 4e15217..cc99347 100644
--- a/cog_safe_push/predict.py
+++ b/cog_safe_push/predict.py
@@ -252,7 +252,7 @@ async def predict(
     inputs: dict,
     timeout_seconds: float,
     prediction_index: int | None = None,
-) -> tuple[Any | None, str | None]:
+) -> tuple[Any | None, str | None, str]:
     prefix = f"[{prediction_index}] " if prediction_index is not None else ""
     log.vv(
         f"{prefix}Running {'training' if train else 'prediction'} with inputs:\n{json.dumps(inputs, indent=2)}"
@@ -283,7 +283,8 @@ async def predict(
             else:
                 raise
 
-    log.v(f"{prefix}Prediction URL: https://replicate.com/p/{prediction.id}")
+    prediction_url = f"https://replicate.com/p/{prediction.id}"
+    log.v(f"{prefix}Prediction URL: {prediction_url}")
 
     while prediction.status not in ["succeeded", "failed", "canceled"]:
         await asyncio.sleep(0.5)
@@ -295,7 +296,7 @@ async def predict(
 
     if prediction.status == "failed":
         log.v(f"{prefix}Got error: {prediction.error}  ({duration:.2f} sec)")
-        return None, prediction.error
+        return None, prediction.error, prediction_url
 
     output = prediction.output
     if _has_output_iterator_array_type(version):
@@ -303,4 +304,4 @@ async def predict(
 
     log.v(f"{prefix}Got output: {truncate(output)}  ({duration:.2f} sec)")
 
-    return output, None
+    return output, None, prediction_url
diff --git a/cog_safe_push/tasks.py b/cog_safe_push/tasks.py
index 7590034..aeb8545 100644
--- a/cog_safe_push/tasks.py
+++ b/cog_safe_push/tasks.py
@@ -28,6 +28,7 @@ class CheckOutputsMatch(Task):
     fuzz_disabled_inputs: list[str]
     fuzz_prompt: str | None
     prediction_index: int | None = None
+    prediction_url: str | None = None
 
     async def run(self) -> None:
         if self.first_test_case_inputs is not None:
@@ -57,7 +58,7 @@ async def run(self) -> None:
         log.v(
             f"{prefix}Checking outputs match between existing version and test version, with inputs: {inputs}"
         )
-        test_output, test_error = await predict(
+        test_output, test_error, test_url = await predict(
             model=self.context.test_model,
             train=self.context.is_train(),
             train_destination=self.context.train_destination,
@@ -65,7 +66,8 @@ async def run(self) -> None:
             timeout_seconds=self.timeout_seconds,
             prediction_index=self.prediction_index,
         )
-        output, error = await predict(
+        self.prediction_url = test_url
+        output, error, _ = await predict(
             model=self.context.model,
             train=self.context.is_train(),
             train_destination=self.context.train_destination,
@@ -97,13 +99,14 @@ class RunTestCase(Task):
     checker: OutputChecker
     predict_timeout: int
     prediction_index: int | None = None
+    prediction_url: str | None = None
 
     async def run(self) -> None:
         prefix = (
             f"[{self.prediction_index}] " if self.prediction_index is not None else ""
         )
         log.v(f"{prefix}Running test case with inputs: {self.inputs}")
-        output, error = await predict(
+        output, error, url = await predict(
             model=self.context.test_model,
             train=self.context.is_train(),
             train_destination=self.context.train_destination,
@@ -111,6 +114,7 @@ async def run(self) -> None:
             timeout_seconds=self.predict_timeout,
             prediction_index=self.prediction_index,
         )
+        self.prediction_url = url
 
         await self.checker(output, error)
 
@@ -150,6 +154,7 @@ class FuzzModel(Task):
     inputs_queue: Queue[dict[str, Any]]
     predict_timeout: int
     prediction_index: int | None = None
+    prediction_url: str | None = None
 
     async def run(self) -> None:
         inputs = await asyncio.wait_for(self.inputs_queue.get(), timeout=60)
@@ -159,7 +164,7 @@ async def run(self) -> None:
         )
         log.v(f"{prefix}Fuzzing with inputs: {inputs}")
         try:
-            output, error = await predict(
+            output, error, url = await predict(
                 model=self.context.test_model,
                 train=self.context.is_train(),
                 train_destination=self.context.train_destination,
@@ -167,6 +172,7 @@ async def run(self) -> None:
                 timeout_seconds=self.predict_timeout,
                 prediction_index=self.prediction_index,
             )
+            self.prediction_url = url
         except PredictionTimeoutError:
             raise FuzzError(f"{prefix}Prediction timed out")
         if error is not None:
diff --git a/test/test_main.py b/test/test_main.py
index 13dec72..413ac84 100644
--- a/test/test_main.py
+++ b/test/test_main.py
@@ -1,13 +1,15 @@
 import pytest
 
 from cog_safe_push import log
-from cog_safe_push.exceptions import ArgumentError
+from cog_safe_push.exceptions import ArgumentError, TestCaseFailedError
 from cog_safe_push.main import (
     parse_args_and_config,
     parse_input_value,
     parse_inputs,
     parse_model,
+    run_tasks,
 )
+from cog_safe_push.tasks import Task
 
 
 def test_parse_args_minimal(monkeypatch):
@@ -476,3 +478,48 @@ def test_parse_args_push_official_model(monkeypatch):
     assert config.model == "user/model"
     assert not no_push
     assert push_official_model
+
+
+@pytest.mark.asyncio
+async def test_run_tasks_reports_all_errors_with_details(capsys):
+    class FailingTask(Task):
+        def __init__(self, prediction_index, prediction_url, error_message):
+            self.prediction_index = prediction_index
+            self.prediction_url = prediction_url
+            self.error_message = error_message
+
+        async def run(self):
+            raise TestCaseFailedError(self.error_message)
+
+    class FailingTaskNoUrl(Task):
+        def __init__(self, prediction_index, error_message):
+            self.prediction_index = prediction_index
+            self.error_message = error_message
+
+        async def run(self):
+            raise TestCaseFailedError(self.error_message)
+
+    tasks = [
+        FailingTask(1, "https://replicate.com/p/abc123", "Output mismatch"),
+        FailingTask(2, "https://replicate.com/p/def456", "Timeout occurred"),
+        FailingTaskNoUrl(3, "Invalid input"),
+    ]
+
+    with pytest.raises(TestCaseFailedError):
+        await run_tasks(tasks, parallel=2)
+
+    captured = capsys.readouterr()
+    error_output = captured.err
+
+    # All errors should be logged
+    assert (
+        "[1] Test case failed: Test case failed: Output mismatch; Prediction URL: https://replicate.com/p/abc123"
+        in error_output
+    )
+    assert (
+        "[2] Test case failed: Test case failed: Timeout occurred; Prediction URL: https://replicate.com/p/def456"
+        in error_output
+    )
+    assert "[3] Test case failed: Test case failed: Invalid input" in error_output
+    # Should not have prediction URL in the third error
+    assert "Invalid input; Prediction URL:" not in error_output