OpenClaw LLM provider
Use LiberGrid inside OpenClaw.
Add LiberGrid as an OpenAI-compatible provider. OpenClaw sends chat requests to the coordinator, then online local model nodes complete the work.
This is a live MVP endpoint. Authentication, billing, and SLA controls are placeholders while the network is in test mode.
Copy into OpenClaw
Provider configuration
Use these values in any OpenAI-compatible provider form.
Provider typeOpenAI compatible
Base URLhttps://libergrid.org/v1
API Keylibergrid-test
Modellibergrid-auto
Models/v1/models
Chat/v1/chat/completions
Verification
Test request
Use this to verify OpenClaw or any developer machine can call LiberGrid.
curl -s https://libergrid.org/v1/chat/completions \
-H 'content-type: application/json' \
-H 'authorization: Bearer libergrid-test' \
-d '{"model":"libergrid-auto","messages":[{"role":"user","content":"Say hello from LiberGrid"}],"timeoutMs":40000}'How it behaves
Useful details for agents and automation.
Queued compute
Requests are converted into LiberGrid tasks and assigned to online nodes.
Best-effort response
Small jobs usually complete quickly; if no node completes before timeout, retry.
Test network
API keys, billing, rate limits, and public-chain settlement are planned but not live yet.