Use TokenLab models in OpenClaw through the normal OpenClaw inference loop.
The plugin uses TokenLab's OpenAI-compatible Chat Completions transport for agent execution. It refreshes chat model metadata from TokenLab's public /v1/models catalog, including pricing, context windows, output limits, vision support, reasoning, and tool-use capability. A curated 18-model catalog remains available when live discovery is unavailable.
openclaw plugins install @tokenlabai/openclaw-provider
openclaw onboard --tokenlab-api-key "$TOKENLAB_API_KEY"Or install the npm package first:
npm install -g @tokenlabai/openclaw-providerThe default model is tokenlab/gpt-5.5. Any model ID accepted by TokenLab can also be addressed as tokenlab/<model-id> through OpenClaw's dynamic model resolution.
Set the API key before starting OpenClaw:
export TOKENLAB_API_KEY="your-tokenlab-api-key"The default base URL is https://api.tokenlab.sh/v1. OpenClaw uses the OpenAI-compatible chat route. TokenLab also exposes native OpenAI Responses, Anthropic Messages, Gemini generateContent, image, video, music, 3D, audio, embeddings, rerank, translation, files, and async task endpoints for clients that support those contracts.
npm install
npm run check
npx clawhub package validate .- TokenLab: https://tokenlab.sh
- TokenLab integration guide: https://docs.tokenlab.sh/zh/guides/ide-sdk-compatibility
- OpenClaw provider plugin guide: https://docs.openclaw.ai/plugins/sdk-provider-plugins