Join Nostr
2026-10-16 20:45:22 UTC

jonny (nonvenomous) on Nostr: I was avoiding reading this because I thought it would be silly, and it is. Top-line ...

I was avoiding reading this because I thought it would be silly, and it is.
Top-line figure that's getting quoted: local LLM can run across 15 TVs at 0.4 tokens/s! Like yes, we already knew sharding layers was possible. Then the rest of it is just like... Nothing?

https://www.anthropic.com/research/self-replicating-ai