My openclaw keeps switching back to Kimi when I tell it to change back to Opus or chatGPT
✅Solved
even restarting the service, my main model is Opus with open AI Codex as fallback THEN Kimi 2.5 and after a restart all sessions are on Kimi ? I asked my main to switch to Opus which it did for 1 sec then switched back to Kimi.
Solution
3) Log line showing the failure (run this, reproduce the flip once, then paste the relevant error)
Common causes after restarts:
- Anthropic creds not available to the service (works in your shell, but launchd/systemd doesn’t inherit it). Fix is usually putting the key/token in
~/.openclaw/.envenv.shellEnv.enabled- Anthropic 429 rate limit / billing backoff, which triggers cooldown + fallback. Docs: https://docs.openclaw.ai/help/faq#why-am-i-seeing-http-429-ratelimiterror-from-anthropic
- Different agent is handling the thread after restart, and that agent doesn’t have Anthropic auth profiles configured (auth is per-agent).
/model statusRelevant doc sections:
- Failover behavior: https://docs.openclaw.ai/help/faq#model-failover-and-all-models-failed
- Switching + checking in-chat: https://docs.openclaw.ai/help/faq#how-do-i-switch-models-on-the-fly-without-restarting
The AI that actually does things. Emails, calendar, home automation - all from your favorite chat app. New shell, ready to help. The lobster way.
134,827Members
Resources
Recent Announcements
Similar Threads
Was this page helpful?
