vitalik.eth @VitalikButerin · 11 Aug 2026
Impressive work! For comparison, an H100 can do roughly 100-200 tok/s of Muse 30B for raw inference single-thread, going up to low thousands of tok/s with a large number of threads - and I am sure that for the massively-multi-threaded case they can optimize the prover further.
302 554Views
1 003Likes
153Reposts
271Replies
25Quotes
130Bookmarks
Is that a lot?
12.8×vs this author's median23 664 views is typical
67Percentile for this authorof 12 recent posts
4.54×vs 1M–10M median66 615 views is typical
3.89%Reachviews ÷ followers
0.48%Engagement rateof viewers reacted