Tokener CA
Service online
Sign up
DocsCommon Errors

Common Errors

Causes and recommended fixes for common errors when supplying Token.

Updated:

failed to start runtime raw chat request

Cause: the local inference framework hangs under concurrent tasks, and the Provider Agent cannot complete the raw chat request.

Recommended fix: change the framework's concurrent task count to 1, then restart the local inference framework and the Provider Agent. If what you see is a similar message such as failed to star runtime raw chat request, handle it the same way.

preflight failed: VISION_CAPABILITY_UNAVAILABLE

Cause: the preflight did not detect the model's image capability, commonly because LM Studio has not actually loaded the target model.

Recommended fix: manually open LM Studio, load the corresponding model, and confirm the local service is running before re-running the Provider Agent preflight or startup flow.