Ask HN: How do you find out if the LLM API is giving degraded responses
A developer on Hacker News asks how others detect degraded responses from LLM APIs like OpenAI, Claude, and Gemini, including slow time-to-first-token, elevated error rates, and model drift. The user …