openclaw - 💡(How to fix) Fix Configured model fallback is not triggered on temporary AI service overload [1 comments, 2 participants]

Summary

When the active model/provider returns a temporary overload error, OpenClaw surfaces the error directly instead of falling back to the configured fallback models.

Observed error:

The AI service is temporarily overloaded. Please try again in a moment.

Expected behavior

If fallback models are configured, transient provider overload / capacity errors should be treated as fallback-eligible conditions.

OpenClaw should attempt the next configured fallback model before returning an error to the user.

Actual behavior

The request fails immediately with the overload message above.

No configured fallback model appears to be attempted.

Why this matters

This is usually temporary provider-side capacity pressure. From the user’s perspective, configured fallback models exist specifically to keep the session usable during this kind of transient provider outage.

Environment

• OpenClaw: v2026.5.4
• Surface: Telegram / main session
• Fallback models: configured, but not triggered for this overload case

Suggested fix

Classify provider overload / temporary capacity errors as fallback-eligible errors.

Potential matching strings / categories:

• temporarily overloaded
• try again in a moment
• provider 429 / 529 / capacity / overloaded style errors

If a fallback chain exists, OpenClaw should continue to the next model before surfacing the failure.