Why did Gemini 2.5 Pro change my model from GPT-4o-mini to Gemini Flash?

I was using the OpenRouter API and had hardcoded model = "gpt-4o-mini" in my code (just copy–pasting, not using a CLI or agent).

While cleaning up, I noticed it was silently changed to gemini-1.5-flash.

I know AI can hallucinate, but this felt more like an instruction or default override. Has anyone else seen this?I was using the OpenRouter API and had hardcoded model = "gpt-4o-mini" in my code (just copy–pasting, not using a CLI or agent).

While cleaning up, I noticed it was silently changed to gemini-1.5-flash.

I know AI can hallucinate, but this felt more like an instruction or default override. Has anyone else seen this?