Skip to content

OpenRouter Claude 3.7 Sonnet (Recommended)

Access multiple models via OpenRouter using unified OpenAI chat format. Supports Claude, GPT, and other models with higher rate limits.

Client Used

ClaudeOpenRouterClient

Environment Setup

Environment Variables
export OPENROUTER_API_KEY="your-openrouter-key"
export OPENROUTER_BASE_URL="https://openrouter.ai/api/v1"  # optional

Configuration

Agent Configuration
main_agent:
  llm: 
    provider_class: "ClaudeOpenRouterClient"
    model_name: "anthropic/claude-3.7-sonnet"  # or openai/gpt-4, etc.
    async_client: true
    temperature: 0.3
    top_p: 0.95
    min_p: 0.0
    top_k: -1
    max_tokens: 32000
    openrouter_api_key: "${oc.env:OPENROUTER_API_KEY,???}"
    openrouter_base_url: "${oc.env:OPENROUTER_BASE_URL,https://openrouter.ai/api/v1}"
    openrouter_provider: "anthropic"  # Force provider, or "" for auto
    disable_cache_control: false
    keep_tool_result: -1
    oai_tool_thinking: false

Usage

Example Command
# Run with Claude 3.7 Sonnet on example dataset
uv run main.py common-benchmark --config_file_name=agent_llm_claude37sonnet output_dir="logs/test"

The agent_llm_claude37sonnet.yaml configuration file provides a ready-to-use setup with the example dataset benchmark.

Benefits vs Direct API

  • Unified chat format
  • Higher rate limits

Documentation Info

Last Updated: September 2025 · Doc Contributor: Team @ MiroMind AI