Constants§
- CACHING_
MODELS - Models that support context caching (min 2048 tokens required) Context caching reduces costs for repeated API calls with similar contexts Reference: https://ai.google.dev/gemini-api/docs/caching
- CODE_
EXECUTION_ MODELS - Models that support code execution (Python) Code execution allows models to write and execute Python code Reference: https://ai.google.dev/gemini-api/docs/code-execution
- DEFAULT_
MODEL - Default model - using stable version for production reliability
- EXTENDED_
THINKING_ MODELS - Models that support Gemini 3 extended thinking levels (minimal, medium)
Only
gemini-3-flash-previewsupportsminimal; stable 3.7/3.8 Flash support low/medium/high with medium as default - GEMINI_
3_ 6_ FLASH - GEMINI_
3_ 7_ FLASH - GEMINI_
3_ 8_ FLASH - GEMINI_
3_ FLASH_ PREVIEW - GEMINI_
3_ PRO_ IMAGE_ PREVIEW - IMAGE_
GENERATION_ MODELS - Models supporting image generation
- REASONING_
MODELS - Models that support thinking/reasoning capability with configurable thinking_level
Based on: https://ai.google.dev/gemini-api/docs/gemini-3
Gemini 3 Pro: supports low, high
Gemini 3 Flash Preview: supports minimal, low, medium, high
Gemini 3.7/3.8 Flash: supports low, medium (default), high —
minimalnot supported on 3.8 and will error - SUPPORTED_
MODELS