Documentation Index
Fetch the complete documentation index at: https://docs.wednesdayai.dev/llms.txt
Use this file to discover all available pages before exploring further.
AI providers
WednesdayAI routes your conversations to any supported AI provider. You choose the model; the gateway handles authentication, retries, and response streaming.Supported providers
Anthropic
Claude models — Sonnet, Opus, Haiku. Supports thinking mode and prompt caching.
OpenAI
GPT-4, o3, and the Codex family. Standard API key or Codex subscription OAuth.
Gemini models via Vertex AI or the AI Studio API key.
Ollama
Run any model locally — Llama, Mistral, Qwen, and hundreds more.
OpenRouter
Unified API gateway. Access 200+ models from a single API key with automatic fallback.
LiteLLM / Bifrost
Proxy-based providers. Route through your own LiteLLM or Bifrost deployment.
Adding a provider
Useopenclaw login to authenticate with a provider interactively:
Routing requests to a provider
In~/.openclaw/openclaw.json, set the default model:
/model command in chat:
Multi-provider routing
You can configure routing rules that send specific agents or channels to different providers. See the Gateway configuration page for the routing table syntax.Checking provider status
openclaw doctor validates credentials for every configured provider and reports which ones are reachable.