Chinese food-delivery company Meituan has open-sourced LongCat-2.0, a Mixture-of-Experts model whose 1.6 trillion total parameters activate roughly 48 billion per token. The system was pretrained on more than 35 trillion tokens across clusters of over 50,000 domestic AI ASICs arranged in superpods, marking one of the largest known training runs completed entirely without GPUs.
WHOA Ok, here it goes. Owl Alpha. V4-scale model, updated DSA, N-gram Embeddings, decent evals, «pre-trained on over 50K AI ASICs … superpods» from *checks notes* Meituan. Yeeeah. these are clearly *not* Ascend 950DT SuperPODs (<<80 GB memory/ASIC). What the hell did they use?!
@teortaxesTex You missed the point. LongCat 2.0 is trained on domestic accelerators
https://longcat.chat/blog/longcat-2.0/
LongCat Sparse Attention plus dedicated long-context data let the model handle 1 million tokens, while task-routed expert groups focus capacity on agentic coding and tool use. Early users on OpenRouter, where the model previously ran anonymously, have already driven it to high call volumes.
Official materials describe performance gains above 35 percent from custom 6D parallelism and memory tweaks but stop short of naming the exact ASIC. Community observers match the specs to Huawei Ascend 910C superpods; independent efficiency numbers or full benchmark tables have not yet appeared.
Users are reacting to Meituan's LongCat-2.0 1.6T MoE model, with positive ones praising its coding performance and long-context features while negative ones criticize inferior metrics versus prior versions and competitors like GLM-5.2.
No Digg Deeper questions have been answered for this story yet.
Damn, that's right. So: "≈48B" (thanks to N-Gram embedding, variable) active, 35T tokens. V4-tier, ≈8e24? Would be the biggest Chinese pretraining on domestic hardware. Some strange "Superpods": > "our accelerators" > "up to 48 machines each" > below 80 GB HBM per accelerator > "The device offers limited HBM bandwidth but a relatively large L2 cache" > "built-in 200 Gbps network adapter within the accelerator"
what is this stuff Was the first truly big Chinese model trained on domestic compute also using some extremely obscure piece of hardware?
@teortaxesTex @zugzwangg_ it said 35T+ tokens
Introducing LongCat-2.0 🐱 1.6T parameters · MoE with ~48B active · 1M context The full model behind Owl Alpha on @OpenRouter — now available.
Built for agentic coding from the ground up: ◆ LongCat Sparse Attention (LSA) — scales efficiently for 1M-context tokens ◆ Zero-Compute Experts — dynamic activation 33B–56B per token, zero wasted compute ◆ MOPD — three specialized expert groups (Agent / Reasoning / Interaction), gate-routed per task
How it stacks up: → Terminal-Bench 2.1: 70.8 → SWE-bench Pro: 59.5 (GPT-5.5: 58.6) → SWE-bench Multilingual: 77.3 → FORTE: 73.2 · RWSearch: 78.8 · BrowseComp: 79.9
📖 Tech Blog: https://longcat.chat/blog/longcat-2.0/ Try it across different scenarios 🧵👇
🎁 Launch offer — limited time: ✦ 50M tokens for just $1.9 ✦ 1B tokens for $59.9 (80% off) Cache hits are free. Unused tokens fully refundable.
→ Try it now: https://longcat.chat/platform/product → Join our Discord for more demos, support, and direct feedback：https://discord.gg/jmuv5TQzKU
Most popular model on @OpenRouter (10tr tokens) turns out to be a 1.6tr MoE by @Meituan_LongCat (superapp/DoorDash of China)
Basically Gemini / Opus 4.6 level
35tr tokens trained entirely on 50k Chinese ASICs
No GPUs needed
https://longcat.chat/blog/longcat-2.0/
Some of you guessed right. 👀 Owl Alpha on @OpenRouter — that's us.
Since going live, it has reached Top 3 globally by daily volume — and #1 on Hermes Agent, #2 on Claude Code, #3 on OpenClaw by monthly volume.
Thank you to everyone who tested and used Owl Alpha during stealth — you helped shape what's coming next. Owl Alpha will be retiring soon. But this isn't an ending — stay tuned!
new 1.6T total, 48B active open model coming very soon from @Meituan_LongCat
custom sparse attention, n-gram, adaptive expert computation, 1M context, 35T of training tokens, muon optimizer, multi teacher OPD, trained on 50k chinese gpus
https://longcat.chat/blog/longcat-2.0/
910B had 64Gb already, and I don't think there exist 910B "superpods". @zephyr_z9 @tphuang I missed these news, they're not even very new. anyway, very cool development on N-grams
tbh as usual very interesting technical details from Meituan but when I click "try it" I get something like Qwen-4B, seemingly worse than their first LongCat-Flash. Completely fails in Russian too. Idk if that's their new model
tbh as usual very interesting technical details from Meituan but when I click "try it" I get something like Qwen-4B, seemingly worse than their first LongCat-Flash. Completely fails in Russian too. Idk if that's their new model
WHOA Ok, here it goes. Owl Alpha. V4-scale model, updated DSA, N-gram Embeddings, decent evals, «pre-trained on over 50K AI ASICs … superpods» from *checks notes* Meituan. Yeeeah. these are clearly *not* Ascend 950DT SuperPODs (<<80 GB memory/ASIC). What the hell did they use?!
Meituan, basically China’s DoorDash, trained a 1.6T parameter LLM on 50K Chinese chips.
It reminds me of Jensen Huang’s point on the Dwarkesh podcast: export controls on Nvidia GPUs won’t stop China. They’ll just accelerate the development of AI that runs on Chinese chips.
Hm, yes, it does match if you remember that 910C is two "910B" dies glued together. Okay then! 384 CloudMatrix – or rather, a hundred or so of them – is good enough to train a 1.6T MoE, it seems. This is 2025 tech. In ≈3 weeks we'll see actual "super"PODs.
@teortaxesTex These are almost definitely Huawei Ascend 910C. 910C superpod has 48 machines with 8 processors each. Each processor has two physical dies that could function as two logical processors. In that mode, each die has 64GB HBM and 200 Gbps RDMA networking so everything matches.
💻 Coding: Reads your entire codebase at once — understands the architecture, tracks cross-file dependencies, and delivers changes that compile clean.
@teortaxesTex These are almost definitely Huawei Ascend 910C. 910C superpod has 48 machines with 8 processors each. Each processor has two physical dies that could function as two logical processors. In that mode, each die has 64GB HBM and 200 Gbps RDMA networking so everything matches.
📊 Analysis: Understands your question in natural language, plans the analysis steps, and delivers business insights — no SQL, no manual wrangling.
DS and Enthropic suspect Enflame 邃思 2.0
Damn, that's right. So: "≈48B" (thanks to N-Gram embedding, variable) active, 35T tokens. V4-tier, ≈8e24? Would be the biggest Chinese pretraining on domestic hardware. Some strange "Superpods": > "our accelerators" > "up to 48 machines each" > below 80 GB HBM per accelerator > "The device offers limited HBM bandwidth but a relatively large L2 cache" > "built-in 200 Gbps network adapter within the accelerator"
what is this stuff Was the first truly big Chinese model trained on domestic compute also using some extremely obscure piece of hardware?
📑 Research: Give it a topic. It searches, reads, cross-references, and refines autonomously until it delivers a structured review.
✍️ Writing: Handles long-form content with consistent structure, tone, and detail throughout, even across rewrites.
Jensen crying in the style of the native american . One single tear
Introducing LongCat-2.0 🐱 1.6T parameters · MoE with ~48B active · 1M context The full model behind Owl Alpha on @OpenRouter — now available.
Built for agentic coding from the ground up: ◆ LongCat Sparse Attention (LSA) — scales efficiently for 1M-context tokens ◆ Zero-Compute Experts — dynamic activation 33B–56B per token, zero wasted compute ◆ MOPD — three specialized expert groups (Agent / Reasoning / Interaction), gate-routed per task
How it stacks up: → Terminal-Bench 2.1: 70.8 → SWE-bench Pro: 59.5 (GPT-5.5: 58.6) → SWE-bench Multilingual: 77.3 → FORTE: 73.2 · RWSearch: 78.8 · BrowseComp: 79.9
📖 Tech Blog: https://longcat.chat/blog/longcat-2.0/ Try it across different scenarios 🧵👇
@zephyr_z9 @tphuang it's strangely identical to V4 in dimensions, and they boast of total determinism too. Replicating V4 in 2 months is a bit rich… but that's not even what happened if LongCat 2.0 is Owl Alpha – it was debuted on OR on Apr 28. What is going on here.
910B had 64Gb already, and I don't think there exist 910B "superpods". @zephyr_z9 @tphuang I missed these news, they're not even very new. anyway, very cool development on N-grams
> For numerical reliability, we rework a range of foundational operators to improve precision — for example, all reduction-type operators adopt a binary-tree segmented accumulation strategy to reduce floating-point error accumulation. We further validate the accelerator's compute precision under real-world LLM workloads against a strict high-precision baseline, confirming its arithmetic integrity and production readiness, and introduce bit-flip detection in selected compute-intensive operators to promptly catch anomalies caused by hardware bit flips.
Sounds like alien techs
For the latest pricing and offer details, please refer to our official website.
@Meituan_LongCat @OpenRouter hey Meituan lets collab, we are building @gitlawb and we are the maintainer of OpenClaude currently 29.5K stars coding cli. we would love to onboard you as provider
