🎁 New Sign up free, 10 calls on us. Up to $1, no card needed.

Tool Compatibility

Which base URL should you paste into your tool? Almost every issue comes down to one of two answers:

  • OpenAI-style tools need https://synthorai.io/v1 (include the /v1).
  • Anthropic-style tools need https://synthorai.io (no /v1 — the SDK appends it).

Dedicated step-by-step guides: Claude Code, OpenAI Codex CLI, and OpenClaw.

Quick reference

ToolBase URL to pasteKey field / env var
Claude Code (CLI)https://synthorai.ioANTHROPIC_API_KEY
Anthropic Python / Node SDKhttps://synthorai.ioapi_key= (constructor)
OpenAI Codex CLIhttps://synthorai.io/v1OPENAI_API_KEY
OpenAI Python / Node SDKhttps://synthorai.io/v1api_key= (constructor)
OpenClaw (provider.type=openai)https://synthorai.io/v1providers.X.api_key
OpenClaw (provider.type=anthropic)https://synthorai.ioproviders.X.api_key
Cursor (OpenAI override)https://synthorai.io/v1Settings → API Key
Continue (config.json)https://synthorai.io/v1models[].apiKey
Aiderhttps://synthorai.io/v1OPENAI_API_KEY
LangChain ChatOpenAIhttps://synthorai.io/v1openai_api_key
LangChain ChatAnthropichttps://synthorai.ioanthropic_api_key
LiteLLM proxysee LiteLLM configapi_key per route
Raw curl (OpenAI-style)https://synthorai.io/v1/...Authorization: Bearer
Raw curl (Anthropic-style)https://synthorai.io/v1/messagesx-api-key header

Which protocol does each tool speak?

The protocol the tool uses determines which upstream models it can reach through the gateway:

  • OpenAI Chat Completions (/v1/chat/completions) — routes to any chat model: OpenAI, Anthropic, Google, Qwen, DeepSeek, and more. This is the universal path — the default for Codex, Aider, Cursor, Continue, OpenClaw-OpenAI, LangChain ChatOpenAI, etc.
  • Anthropic Messages (/v1/messages) — routes only to Claude-family models via Anthropic or Bedrock-Anthropic channels. Used by Claude Code, the Anthropic SDKs, OpenClaw-Anthropic, and LangChain ChatAnthropic.

If a tool isn't listed, try the OpenAI shape first (https://synthorai.io/v1 + any chat model). That covers 90% of community integrations.

Once your tool is wired up, pick a default model from the model price comparison — or estimate what a month of usage costs with the LLM cost calculator.