- Log gpt_provider and model in preflight info - Return structured HTTP 503 with actionable error details - Include available_providers, requested_provider, and suggestion - Help users understand what went wrong and how to fix it
26 KiB
26 KiB