Overview
Plug models, tools, and agent harnesses into BitRouter.
Every integration is the same move: point something at BitRouter's endpoint (local proxy at http://127.0.0.1:4356, or Cloud at https://api.bitrouter.ai) and address models by their provider/model id. From there you get the whole registry, provider selection, and fallback underneath. New to BitRouter? Start with the Quick Start.
Integrations come in three kinds:
Models
Where your tokens come from — a subscription you already pay for, an aggregator key, or a model you serve yourself.
Models overview
The three shapes of model source
Claude subscription
Route your Claude Pro/Max plan — OAuth, no key
Codex subscription
Route your ChatGPT plan via the Codex backend
OpenRouter
Bring an OpenRouter key
Ollama
Run open models locally · :11434
vLLM
High-throughput GPU serving · :8000
Tools
Web-search and retrieval providers your models call through BitRouter's built-in web search.
Tools overview
How search providers back the web_search tool
Exa
Neural / semantic web search
Parallel
Web search + research API for agents
Firecrawl
Search + scrape to LLM-ready markdown
Tavily
Search built for agents and RAG
Harnesses
Agent runtimes that drive their own loop — point one at BitRouter and run it on any model.
Harnesses overview
Harness vs. model source
Claude Code
Anthropic Messages via ANTHROPIC_BASE_URL
Codex
Custom provider in ~/.codex/config.toml
OpenCode
Provider block in opencode.json
Hermes
Nous Research's self-improving agent
OpenClaw
Messaging gateway across many channels
Switching gateways
How is this guide?