Hacker News Reader

So it's back to the original question, why spend $5-10k on the Studio, when it will still be 10x slower and half the intelligence vs. $20 Sonnet?.. What is the point (besides privacy) to use local models now for coding?

PS: I can understand that isolated "valuable" problems like sorting photo collection or feeding a cat via ESPHome can be solved with local models.

View on HN

At least for me, it's cheap. Even Claude Haiku 4.5 would cost over $60 each day for the same token amount, after accounting for electricity costs. I have the hardware for other reasons anyway, so why not use it, avoid privacy issues and save money.

Are the LLMs very useful? That is a whole other discussion...

You can't use a $20 Sonnet subscription for general agentic use cases, you have to pay for API use on a per-token basis. The $20 and $200 subscriptions are widely considered unsustainable as such. If anything, the real competition is third-party cheap inference providers.

Reply on news.ycombinator.com