OpenClaw keep replying with error code
✅Solved
I've set up for telegram and starting using it, but then I ask if i can send images in telegram and it could read it, it says that i can try, I tried, its just now sending this error "400 model "claude-sonnet-4.5" not supported for vision. no matter what i type now its just replying that. how to fix this, anyone?
Solution
Yep — you can stop the error without losing the session, and you can also make it not happen again.
In the same Telegram chat, send as a standalone message:
-
(If you have an “Opus” option in the picker, that’s usually the easy Anthropic vision choice.)
After that, your existing session should work again.
-
Keep the session (no reset)
Your session is “stuck” because there’s an image in the session history, and your current model can’t accept images. Fix is: switch the session model to a vision-capable one.In the same Telegram chat, send as a standalone message:
-
/model(If you have an “Opus” option in the picker, that’s usually the easy Anthropic vision choice.)
After that, your existing session should work again.
Go back to your text model and avoid re-triggering the image
Once you’re unstuck, run:-
/compact
The AI that actually does things. Emails, calendar, home automation - all from your favorite chat app. New shell, ready to help. The lobster way.
83,815Members
Resources
Recent Announcements
Similar Threads
Was this page helpful?
