pick_model.py is the truest sentence written about LLM apps this year. Zero markup, MIT, BYOK and no Postgres = chef’s kiss. This is how infra should ship. model="auto" picking the cheapest capable model is the abstraction we’ve all been hand rolling forever. The cross provider deterministic cache is gold, boring until your bill drops hard. Unlike OpenRouter, this one doesn’t take a cut. Just clean and honest routing. Huge appreciation for OrcaRouter-Lite. This is excellent work. Star it if you ship LLMs

@future_coded The fact that it just works with docker compose up

@future_coded Amazing share

@future_coded That cross-provider prompt cache is going to save some money

@future_coded Mate, every LLM app has that cursed file somewhere

@future_coded Great opportunity

@future_coded Excellent

@future_coded Great tool

@future_coded Great idea

@future_coded That's Amazing share

@future_coded This is what developer friendly actually means

@future_coded This is powerful

@future_coded Prompt cache that actually works and doesn’t cost extra? I’m sold

@future_coded No Postgres, no Redis, just works. My kind of tool

@future_coded Excellent share

@future_coded Thank you for sharing your insights on OrcaRouter-Lite. It’s refreshing to see an approach that emphasizes efficiency and cost-effectiveness in LLM applications. I appreciate the emphasis on transparency in routing. Well done!

@future_coded OrcaRouter-Lite looks excellent

@future_coded Great tool

@future_coded feels like open source from 2015 in the best way

@future_coded Impressive share

@future_coded Excellent

@future_coded Finally someone shipping infra without trying to tax

@future_coded This is gold mate

@future_coded Great tool

@future_coded Dropping everything to try this right now, looks insane