Comment by westurner

Gemini, for example, launched implicit caching on or about 2025-05-08: https://news.ycombinator.com/item?id=43939774 re: same:

> Does this make it appear that the LLM's responses converge on one answer when actually it's just caching?