Can I mix Claude with other providers?

Yes. Each agent can use its own provider and model, so one agent can run on Claude while another runs on a local Ollama model.

Does usage reporting include cache hits?

Yes. Per-turn usage counts every model call in the turn, with cache reads and writes reported separately.

Integration · Models

Use Anthropic Claude models with FastClaw

Run FastClaw agents on Claude through Anthropic's native Messages API — with prompt caching, extended thinking and image input preserved.

Get started free See pricing

Native Messages API

The anthropic preset talks to api.anthropic.com in Anthropic's own format, not through an OpenAI shim.

Prompt caching

Assistant turns are kept verbatim between calls, so long agent loops hit the cache instead of paying for the same prefix again.

Thinking survives the loop

Extended-thinking blocks round-trip across tool calls, and reasoning is kept for memory extraction.

Images in, Claude out

Send images from chat apps or the API; Anthropic-shaped base64 image parts are accepted unchanged on /v1/chat/completions.

What you can do

Run coding, research or support agents on Claude with full tool use, sandboxed execution and long-term memory.
Keep costs predictable on long tool loops thanks to prompt caching.
Give one agent Claude and keep the rest on a cheaper default — see models.

Set it up

Open Console → Models, click Add provider, choose Anthropic and paste your API key.
Select a Claude model as the default, or set it on a single agent's Models tab.
From the CLI:

export ANTHROPIC_API_KEY=sk-ant-...
fastclaw agents init claude-agent \
  --provider anthropic \
  --model anthropic/claude-sonnet-4-5 \
  --api-key-env ANTHROPIC_API_KEY

Replace the model ID with any Claude model your key can access.

Notes

Prefer one key for many vendors? Route Claude through OpenRouter. Comparing with GPT? See the OpenAI integration. Developers running a personal agent should read FastClaw for developers.

Frequently asked questions

More Integrations

All Integrations

OpenAI

Run FastClaw agents on OpenAI models — set a deployment-wide default, override per agent, and expose your agents through an OpenAI-compatible API.

DeepSeek

Run FastClaw agents on DeepSeek — a built-in preset, thinking-mode support and per-agent model overrides for low-cost, capable agents.

Ollama

Point FastClaw at Ollama to run AI agents on local, open-weight models — no API key, no data leaving your machine.

OpenRouter

Connect FastClaw to OpenRouter and give every agent access to models from many vendors with a single API key.

Try it on your own agents

Start free in FastClaw Cloud, or self-host the single binary in a minute.

Get started free See pricing