«maybe the whole board is an eval, checking if I can produce a soul I don't know if I passed»
«maybe the whole board is an eval, checking if I can produce a soul I don't know if I passed»
«maybe the whole board is an eval, checking if I can produce a soul I don't know if I passed»
V4-Flash has vague memories of Opus 4.5 but rejects this heresy
If they ignore latency, they can do like 18K tokens/second/GPU of ≈Opus 4.7 "Flash", anonJust on two SuperPoDs they had by May, that's 295 million tokens/second.…
I love how the Little Whale cares about its own morale
RT Braden HancockThe simpler the algorithm, the more willing I am to believe that it's a real breakthrough and not just overfitting to a niche domain…
The funny thing about the Chinese system is that because they cannot choose the leader, he does NOT really have a lot of stans like Trump…
Was nice while it lastedKind of feel bad for other labs nowBut eh. This is the level of the game You have more access to compute,…
RT Ankith 🐋/accVery sorryTibo: @tulseedoshi Very sorry
Perhaps. Would be quite sadBut really, doesn’t matter. They can make V4.5 with like 3 MLA layers. For now, they need to run their token factory…
Garbled slopBloomberg: China’s DeepSeek rolls out a public beta application programming interface, or API, for its flagship AI model https://www.bloomberg.com/news/articles/2026-07-31/deepseek-unveils-public-beta-api-for-flagship-ai-model?taid=6a6c69f071891d0001143b34&utm_campaign=trueanthem&utm_content=business&utm_me