Use Claude with OpenClaw
$3/Mtok inOpenClaw agents run long sessions unattended — Sonnet's reliability under tool use is what you want steering them, with per-agent keys capping the downside.
Facts
At a glance
Model
Claude Sonnet 4.5
anthropic/claude-sonnet-4-5
Price
$3 / M input · $15 / M output
Context window
200k tokens
Tool
OpenClaw
Setup below — under two minutes
Quickstart
Set up Claude in OpenClaw
- 1
Set the agent runtime's LLM endpoint:
ANTHROPIC_BASE_URL=https://api.companyfabric.com ANTHROPIC_API_KEY=<per-agent CompanyFabric key> - 2
Set the agent's model to anthropic/claude-sonnet-4-5 and give the key a budget cap — runaway agents stop at the cap, not at the invoice.
Full tool guide: Use any model with OpenClaw agents →
FAQ
Frequently asked questions
Do I need a Anthropic account to use Claude in OpenClaw?
No — one CompanyFabric key covers Claude Sonnet 4.5 and every other model in the library. Billing is one prepaid balance with exact per-call metering; BYOK at 0% is available if you already have provider keys.
What does Claude Sonnet 4.5 cost in OpenClaw?
$3 / M input · $15 / M output — the same metered price as any other route to the model. Set a per-key budget cap to bound what the tool can spend.
Can I switch models later without touching OpenClaw again?
Yes — the endpoint stays the same; the model is a string. Any catalog id works in the same configuration, including cheaper tiers for bulk work.
Explore
Related setups
Use Claude with Cursor
Cursor's UX with Claude Sonnet as the brain — the pairing most teams standardize on. Routing it through one metered key adds per-seat spend visibility Cursor's own billing doesn't give you.
Use Claude with Cline
Cline's act loop was practically designed around Claude's tool use. One key also lets you route Cline's plan mode to a cheaper tier and keep Sonnet for execution.
Use Claude with Continue
Continue's role system pairs naturally with the library: Sonnet on chat/edit, a fast tier on autocomplete — one config block, one balance.
Use Claude with Aider
Aider's benchmark-driven culture keeps landing on Claude for the editor model. Route it through one key and pair a cheap weak model for commit messages and repo maps.
One key. Every model. Exact prices.
Run Claude in OpenClaw with free starter credits. Free starter credits included — no subscription required.