News for hackers. Stuff that matters.
Re:Astral to Join OpenAI by palmotea on Thursday March 19, 2026 @10:40AM (#47440347)
> you need enormous VRAM laden farms of GPUs to do inference on a model like Opus 4.6.

It's probably a trade secret, but what's the actual per-user resource requirement to run the model?

There are no comments on this comment yet.