Model request failed
Symptoms
The client connects, but a request fails, returns an error, or does not produce the expected model response.
Check
- Open Request history and inspect the newest matching interaction and visible failure stage.
- Confirm the selected Stravia model has a valid route and the API key permits it.
- Check the destination service's availability, credentials, and upstream response.
- Check Allowance for the provider's current quota snapshot. Note that allowance values are stale, unknown, or error states depending on provider support; they are not enforced spend caps. Reported usage is controlled separately by the upstream provider.
- Check whether the interaction used a tool, media input, or plugin whose own configuration can fail independently.
Resolution
Fix the stage indicated by the request evidence, then issue a new controlled test request. Do not infer that a copy/apply action succeeded, or that a recorded interaction means the upstream model answered successfully. Avoid repeating a request if the upstream outcome is uncertain.
Escalate if still stuck
Share a redacted request stage/error, model name, route, and service provider. Never share raw keys, prompts, attachments, or unreviewed debug data. See Request history and Diagnostics .