Skip to main content

​
Setup

  1. Get an API key from platform.openai.com.
  2. Set the environment variable:
export OPENAI_API_KEY="sk-..."

​
Configuration

agents:
  root:
    model: openai/gpt-4o

​
Available models

ModelBest forNotes
gpt-5Most capable, complex reasoningHighest capability
gpt-5-miniFast, cost-effective, good reasoningRecommended default
gpt-4oMultimodal, balanced performanceVision support
gpt-4o-miniSimple tasks, lowest costFastest
For a complete and up-to-date list, see modelname.ai.

​
Thinking budget

OpenAI reasoning models support an effort level string: 
models:
  gpt-thinking:
    provider: openai
    model: gpt-5-mini
    thinking_budget: medium  # minimal | low | medium (default) | high
To disable thinking: 
models:
  gpt:
    provider: openai
    model: gpt-5-mini
    thinking_budget: none

​
Custom endpoint

Use base_url to connect to OpenAI-compatible APIs such as proxies or Azure OpenAI: 
models:
  proxied:
    provider: openai
    model: gpt-4o
    base_url: https://your-proxy.example.com/v1
    token_key: PROXY_API_KEY  # optional if auth is required
For reusable proxy configuration across multiple models, define a custom provider in the providers section. See Custom providers.

Build docs developers (and LLMs) love

Get started for freeTalk to us