Use OpenClaw with Manifest
OpenClaw is the open source agentic CLI for terminal-driven coding. Pair it with Manifest and every prompt gets routed to the cheapest capable model, cutting your AI bill by up to 70% without changing how you code.
— models routed through Manifest over the last 30 days.
Why use OpenClaw with Manifest
OpenClaw burns through tokens fast: long agentic sessions, large context windows, constant tool calls. Manifest sits in front of OpenClaw's API endpoint, inspects every request, and picks the smartest model for the job. Cheap models for boilerplate, premium models for hard reasoning.
- Cut API costs by up to 70% on a typical OpenClaw workload
- Keep using the OpenClaw CLI you already know, no workflow change
- Mix Anthropic, OpenAI, Google, and open models in one routing rule
- Set monthly spend limits and alerts per agent
- Track token usage, latency, and cost per session from the dashboard
- Open source, self-hostable, no lock-in
Setup in 60 seconds
Register Manifest as a provider in your OpenClaw config and point the default model at
manifest/auto. Pick CLI commands or the
built-in openclaw onboard wizard. Both
land at the same result.
Register Manifest in your OpenClaw config to route each request to the best
provider using the model
manifest/auto.
Set the provider config and default model directly via CLI commands.
openclaw config set models.providers.manifest '{"baseUrl":"https://api.manifest.build/v1","api":"openai-completions","apiKey":"mnfst_YOUR_KEY","models":[{"id":"auto","name":"Manifest Auto"}]}'
openclaw config set agents.defaults.model.primary manifest/auto
openclaw gateway restart
Run the onboarding wizard and select Custom Provider when prompted. Then enter the following values:
openclaw onboard
https://api.manifest.build/v1
mnfst_YOUR_KEY
OpenAI Chat Completions-compatible
auto
manifest
OpenClaw tokens processed on Manifest
Models in use
# Model |
Auth type | Tokens / last 30 days |
|---|---|---|