Setup
- Get an API key from aistudio.google.com.
- Set the environment variable:
export GOOGLE_API_KEY="AI..."
Configuration
agents:
root:
model: google/gemini-2.5-flash
models:
gemini:
provider: google
model: gemini-2.5-flash
temperature: 0.5
agents:
root:
model: gemini
Available models
| Model | Best for |
|---|
gemini-3-pro | Most capable Gemini model |
gemini-3-flash | Fast, efficient, good balance |
gemini-2.5-flash | Fast inference, cost-effective |
gemini-2.5-pro | Strong reasoning, large context |
Thinking budget
Thinking budget configuration differs between Gemini 2.5 and Gemini 3 model families.
Gemini 2.5 uses token-based budgets (integers). Gemini 3 uses level-based budgets (strings). Use the correct format for your model version.
Gemini 2.5 (token-based)
Gemini 3 (level-based)
models:
gemini-no-thinking:
provider: google
model: gemini-2.5-flash
thinking_budget: 0 # disable thinking
gemini-dynamic:
provider: google
model: gemini-2.5-flash
thinking_budget: -1 # dynamic — model decides (default)
gemini-fixed:
provider: google
model: gemini-2.5-flash
thinking_budget: 8192 # fixed token budget
models:
gemini-3-pro:
provider: google
model: gemini-3-pro
thinking_budget: high # default for Pro: low | high
gemini-3-flash:
provider: google
model: gemini-3-flash
thinking_budget: medium # default for Flash: minimal | low | medium | high