RT Google GemmaGemma 4 just hit 200M downloads in only 2.5 months!For context, total downloads across the entire Gemma family of models were at 100M when…
RT Google GemmaGemma 4 just hit 200M downloads in only 2.5 months!For context, total downloads across the entire Gemma family of models were at 100M when…
man, do I loathe Bentham's bulldog he's correct on the capabilities issue (it's easier to produce virtuous AIs than reliably appoint virtuous humans to positions of…
RT BraintrustWe analyzed 1,781 real agent traces from @huggingface to understand what actually drives agent success across models, benchmarks, and harnesses.What we found:- The harness matters…
RT Sergio Paniegoyou can now train @liquidai's LFM2-VL in TRLGRPO and RLOO included, with an example script
Griffin and Xavier 20 years agoMaye Musk: 20 years ago at @SpaceX Two photos with @ToscaMusk , @elonmusk and twins.Don’t you also wish you had a…
Tbh I'm a bit disappointed by Huawei. They have now reached ≈Hopper class GPUs. On networking sophistication, 950 generation might exceed *Vera Rubin*, so it will…
This is the beginning of serious autoresearch that can't be withdrawn with the push of a button by some responsible AI safety committee. No matter what…
This is a fascinating and important set of data which shows us where things are going, using OpenAI as a canary in the coal mine.The chatbot…
RT Revant Teotia🎉 Our paper MJEPA has been accepted at #ECCV2026 !!Huge thanks to my awesome collaborators @AdrienBardes, @michaelrabbat , Sumit Chopra @mattmucklm and Nicolas BallasPaper:…
Does this answer your question about its positioning @TheZvi ? I told you it's a hardware issue. GLM 5.2 can be served very, very quickly. Databricks…
RT Niels RoggeImpressive release by @ornith_!Getting the top spot on Terminal Bench 2.1 with only GLM-5.2 (a much bigger model) above it!Also beating Opus 4.8 👀Ornith:…
BigDmytro Dzhulgakov: bit-equivalent on-policy rl for glm-5.2 has been achieved internallydeveloping…
Welcome to Open Source AI: Run Your Own Models Locally https://x.com/i/broadcasts/1AGRnnELNODGl
Opus to Gemini:徐樂 xule: opus4.8 on working with kimi (k2.6), codex (gpt5.5), and gemini (3.5 flash)
RT Xiuyu LiCode Agent, Generalist Agent 和更贵的 Agent 数据DeepSeek 寻找闪亮发光的你 https://mp.weixin.qq.com/s/LFtjgHJpDt-TPfjD0oUM5w永雏塔菲: deepseek 看来是要猛猛发力了，在现有团队规模基础上至少扩大一倍😲
> what the hell does "shtel" even mean and how do you know that?If you're in tech and you don't know what "shtetl" is, you're ngmiyeah…
RT Planet Of MemesThis. 👇 👇 👇zerohedge: *JUDGE BLOCKS TRUMP’S ORDER TO IMPOSE NEW LIMITS ON MAIL VOTINGthat was quick
RT Sawyer MerrittTesla has announced that it will increase production output at its Giga Berlin factory by another 20% to 7,500 vehicles per week starting in…
insane actually how much Anthropic has accelerated in 2026open source models are so often within striking distance of GPT 5.5, but Opus 4.8 is consistently an…
The AI companies should be presenting this much more as transparent data of what is happening over time and much less like smear campaigns with strong…
RT François CholletAgentic coding forces you to design clean interfaces and document them well. An agent cannot read the implicit mental model shared by your engineering…
«One of my personal benchmarks of ASI is a model which recognises that the civilisation of martin luther is different from civilisation of martin luther jr…
RT Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)Re Translation of @deepseek_ai's entire recruitment post "At least double all departments" implies more than doubling the whole org, I…