TokelyConnect OpenClaw to Claude, GPT, Kimi, DeepSeek and Gemini through Tokely. Add a custom provider with the OpenAI-compatible or Anthropic-compatible endpoint.
By TokelyReviewed In OpenClaw, run openclaw onboard, choose Custom Provider and enter https://api.tokely.me/v1 with OpenAI-compatible compatibility, your Tokely key and a model ID. Claude models can also use https://api.tokely.me with Anthropic-compatible compatibility. OpenClaw checks the endpoint before saving it.
# 1. Start the guided setup
openclaw onboard
# Quick start -> More... -> Custom Provider
# 2. OpenAI-compatible (any Tokely chat model)
Base URL: https://api.tokely.me/v1
API key: sk-your-tokely-key
Compatibility: OpenAI-compatible
Model ID: kimi-k2.7-code (or claude-sonnet-5, deepseek-v4-pro-0813, gpt-5.4)
# Claude models can also use the native Anthropic endpoint:
Base URL: https://api.tokely.me
Compatibility: Anthropic-compatible
Model ID: claude-sonnet-5Base URL https://api.tokely.me/v1. Create a key in the dashboard and keep it in an environment variable. Quickstart · Model IDs and endpoints
One key and balance cover every model, so you can switch OpenClaw between Claude, GPT, Kimi K2.7 Code and DeepSeek by changing the model ID. Prompt caching lowers the cost of long agent sessions, and a separate API key with its own budget caps what an assistant can spend.
OpenAI-compatible works for every Tokely chat model. Anthropic-compatible uses the native /v1/messages endpoint and works for Claude models only. To change the provider later, run openclaw onboard again.
Claude Sonnet for the strongest agent behavior, Kimi K2.7 Code or DeepSeek V4 Pro for lower cost. Compare cost per finished task on your own workload.
Check the model, compare the rate, and connect with a Tokely key.