> ByteDance: installingcacklingbut yeah! We're still living in the Hopper era. + TPUs/Trainium. "Blackwell class models" like Mythos or GPT-5.x have virtually nothing to do with…
> serve at DeepSeek-API-killing prices. That loop is why xAI runs Grok on SGLang and third parties> beat DeepSeek's own API by 5x on cost.what is…
RT サメQCUhttps://x.com/voooooogel/status/2074217167600734647"unfortunately it's still not perfect, the main limitation imo being the reliance on the vocabulary. "personally im kind of tired of the categorical over tokens…
Very important whether these assumptions hold.will depue: A Stargate for DataLabs are on a trajectory towards >$100B/year of data spend by 2030. As we begin the…
Fascinating toolon-the-fly model surgeryThat's the kind of tool that I imagine underpins Anthropic's science of pretraining which elevates them from the industry-standard alchemy to serious physics…
Finally a new bench where GLM-5.2 crashes to mid-tier modelsArtificial Analysis: Announcing AutomationBench-AA, our independent leaderboard for Zapier’s AutomationBench, testing whether AI agents can automate real…
Convincing.Anthropic: For most things, Claude actually doesn’t need its J-space. If we delete the J-space, Claude still speaks fluently, recalls facts, and classifies text—but becomes bad…
Fascinating specimen that spells out what is often quietly assumedDr. Bunsen Honeydew: @Bluebearmonkey In wartime, yes. In peacetime, the opposite is better because you can always…
RT Philip TurnerThe profit motive that won the last venture capital round is falling apart.Since this is taking longer than expected, I'm returning to STM development,…
> resumed after 318h58m0s idle (provider cache expired) — elided 1 stale tool results to cheapen the cold restartI like this new feature of Reasonixcache-centric agents