LongCat-2.0 is now live on DGrid ⚡️ Built for real-world Agentic Coding: ▸ 1.6T parameters, ~48B activated ▸ Efficient 1M-token long-context inference ▸ LongCat Sparse Attention for faster long-context workflows ▸ N-gram Embedding + dynamic activation for stronger code understanding, generation, and execution ▸ Agent, reasoning, and interaction capabilities in one model ▸ Open weights with BF16 / FP8 / INT8 deployment options Excited to bring @Meituan_LongCat to more builders on DGrid. Build, route, and scale your agents with LongCat-2.0. #AI# #DGrid#

Longcat-2.0 vs GPT-5.5 (xhigh) Task: Generate a single self-contained HTML file for each of these 6 animations, no external libraries or dependencies: • Glass cracking 🪞 • Dominos falling 🁢 • Building collapsing 🏢 • Tree changing through seasons 🌳 • Lava lamp 🌋 • Snow globe shake ☃️ Longcat-2.0 delivered results on par with GPT-5.5 (and better on a few), while using significantly fewer tokens.

Chinese AI labs are catching up quickly, with recently released GLM-5.2 and Meituan releasing LongCat-2.0 (trained and run entirely on domestic chips). Recently, a group of U.S. AI researchers traveled to China to get a firsthand look at the country’s fast-moving AI ecosystem. During the trip, they visited AI labs and companies across Beijing, Hangzhou, and Shanghai, meeting with teams from Alibaba, Moonshot AI, Zhipu AI, Tsinghua University, Meituan, Xiaomi, Qwen, Ant Group, and Nathan Lambert @natolambert , a prominent AI researcher known for his work on RLHF and open-source AI, joined the trip. What did someone who has spent years at the front lines of America’s open-model movement see when he stepped inside China’s AI labs? How is the competition between the United States and China changing? And what does the next phase of the global AI race look like? In this episode of Silicon Valley 101, we flew to Seattle and sit down with Nathan to talk about what he learned in China — and where he believes the AI world is heading next. English version: Chinese subtitle version:

Latest open artifacts (#20#): New orgs! New types of models! With Nemotron Super, Sarvam, Cohere Transcribe, & others The top end of the market was quiet, but "industry-scale tinkering" just got very loud. We're seeing a massive shift: specialized, cheap open models are now the crucial tools complementing closed agents. Analysis & breakdowns of: - @NVIDIAAI Nemotron-3-Super-120B: 1M context + NVFP4. - @cohere Transcribe - @SarvamAI -105B - @MistralAI -Small-4-119B Massive coverage of new artifacts from: - Coding & Logic: @zeddotdev @Meituan_LongCat & Goedel-LM - Multimodal/OCR: @Microsoft @YuanAI_Lab @BaiduAI & @xiaohongshu (RedNote) - RAG & Search: @trychroma @LightOnIO & @miromind_ai - Robotics/Agents: @Reka_AI @hcompany_ai & @ServiceNowRSRCH - Creative & Audio: @bfl_ml @hume_ai & GAIR - Data & Infra: @allen_ai @IBM @markov__ai @StepFun_ai By @natolambert & @xeophon