Hey everyone!
I’m trying to get my "Allowed/Ignored" provider restrictions and fallback setup working correctly in OpenRouter:
Case 1: Enforced Providers Not Respected
• In my account settings, I exclusively enabled only “Anthropic” and “Google AI Studio” as Allowed Providers with “Always enforce” ON.
• My primary model was set to “openai/gpt-4o-mini”, with fallbacks defined as “anthropic/claude-3.5-sonnet” and “google/gemini-flash-1.5”.
• Despite the enforced selection, OpenRouter routed to the primary model via Azure, which was not on my approved list. I expected requests to skip the unavailable primary provider/model and proceed to the fallback models on allowed providers only.
Case 2: Ignored Provider Still Used
• In another test, I listed “Anthropic” as an Ignored Providers in account settings, with “claude-3.5-sonnet” as the primary model and other fallback models specified.
• OpenRouter still successfully routed to “claude-3.5-sonnet” via Anthropic, ignoring the exclusion, and did not trigger the fallback chain.
Am I missing a config step, or is this a bug? Has anyone figured out how to get strict provider enforcement working for both normal and fallback requests?
I will greatly appreciate if anyone can help me figure this out! thanks