Skip to content

Comment on Show HN: liteLLM Proxy Server: 50+ LLM Models, Error Handling, Cachingparent

Comments

I do think this is an apt analogy. I've heard a counterpoint that there won't be enough "destinations" for this to work, but then it's not hard to imagine a single order of magnitude more LLM "destinations" than today (all of which were launched in the last 12 months).

There's also the fact that the data being sent to these LLM "destinations" could be significantly more valuable (or contain significantly more sensitive information) than the average segment identity or track objects.

AboutSource Built by g1lg1l

Hackerly is an independent reader for Hacker News, built on the public HN API. Not affiliated with Y Combinator.