Model request failed

Symptoms

The client connects, but a request fails, returns an error, or does not produce the expected model response.

Check

  1. Open Request history and inspect the newest matching interaction and visible failure stage.
  2. Confirm the selected Stravia model has a valid route and the API key permits it.
  3. Check the destination service's availability, credentials, and upstream response.
  4. 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.
  5. 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 .