Default model `gemini-3-pro-preview` hits quota limits frequently - Consider switching to gemini-2.5-flash

[omsherikar]

Environment

macOS 26.2 ,Bun v1.3.8, Node.js v24.2.0, pnpm 10.16.1

Steps to Reproduce

1. Clone BubbleLab repo
2. Set up environment variables (GOOGLE_API_KEY)
3. Run pnpm run dev
4. Try to generate a workflow
5. Get quota error: [429 Too Many Requests]

Expected Behavior

• Should use a model with better availability (e.g., gemini-2.5-flash)
• OR implement automatic fallback to alternative models
• OR document quota requirements prominently

Actual Behavior

• Defaults to gemini-3-pro-preview in multiple places
• No fallback or retry logic
• No documentation warning about quota limits

Relevant Logs/Console Output

Screenshots

No response

BubbleLab Version

v0.1.92

Additional Context

No response

[omsherikar]

Hey @zhubzy, we can do

Proposed Solutions

Switch default model to gemini-2.5-flash - it's cheaper, more available, and works just as well.

Better solution: Add automatic fallback so if one model fails, it tries another (gemini-3-pro → gemini-2.5-flash → claude).

Alternative: Just warn people in the README about quota limits and how to pick a different model.