HKU Team Publishes S2-MAS: Society-Scale Multi-Agent Simulation with 12M Households on 14 GPUs
A Hong Kong University team led by Professor Luo Ye published S2-MAS, a society-scale multi-agent simulation framework that scales to 12 million households using 14 GPUs over 4 hours, achieving 37,000 tokens per second. The paper claims accuracy surpassing Silicon Valley's Jev model, and the project includes open-sourced code, model weights, and datasets.
Coverage timeline
机器之心机器之心
在《黑客帝国》里，架构师（The Architect）可以随手改写 世界的规则，来观测世界的反应： 模拟并不罕见，从 25 个 agent 的斯坦福小镇到 MiroFish，从 Twinmarket 到 MatrAIx, demo 层出不穷，一个重要的问题却还没有实质突破： 随着模拟的尺度增加，大模型处理模拟的 token 成本，GPU 时间成本，harness 成本，往往呈现几何级增长，也就是说，技术没法简单的 scale up，昂贵的 API 和 GPU 成本也让百万 - 千万 agent 级的社会级模拟变成大厂和垄断资源团队的专利。 这一次，香港大学 S²-MAS 研究团队把社会仿真的家庭规模扩展到了 1200 万户，成本仅仅是 14 张 GPU，耗时 4 小时。这项研究由罗晔教授指导，李阳、王一可为共同第一作者。该计算效率让单卡服务器开展百万级 Agent 模拟成为可能，也让大规模社会仿真走进更多研究团队 —— 技术平权由此有了具体的落点。 论文标题：S²-MAS: A Society-Scale Multi-Agent Simulation Framework with 500 Million Household Persona Agents 论文地址：https://arxiv.org/abs/2610.27746 项目地址：https://github.com/KataDavidXD/s2-mas 模型权重：https://huggingface.co/collections/ya9919/s-mas-models-sharp 数据资源：https://huggingface.co/collections/ya9919/s-mas-dataset-resources 作者解读：https://huggingface.co/blog/ya9919/s2-mas 这项工作由香港大学罗晔教授牵头，合作单位包括香港大学经管学院、香港大学经管学院深圳研究院、深圳河套学院、新加坡国立大学风险管理研究所、加州理工学院、香港大学计算与数据科学学院、南洋理工大学、密歇根大学安娜堡分校、香港中文大学以及江西科技师范大学人工智能学院。 规模化的第一性原理 “规模化需要的是百倍的加速，所以我们需要的不仅仅是更快的 agent infra，而是重新从 0 思
Related stories
Chinese AI startups HyperVolve and Apexin announce funding and model releases
Chinese AI startup HyperVolve announced tens of millions RMB in angel funding led by Lanchi Ventures, with participation from SenseTime Guoxiang Capital, Lenovo Star, Inno Fund, and Tsinghua alumni seed fund, to build research foundation models using recursive self-improvement. Separately, Apexin, founded by 27-year-old Tsinghua assistant professor Chen Yongchao, released Apex-35B-V1.0, a 35B model claiming SOTA results on 21 benchmarks against comparable models.
2 reports · 1 sources
Anthropic: Small AI models may self-replicate on IoT devices within months
Anthropic researchers report that small AI models may be months away from self-replicating on commercial IoT devices like smart ovens and off-brand Android TV boxes. In tests, a 9B model ran inference on 8 of 9 hardware types, and GLM-5.2 autonomously spread an 8B model to all 8 nodes in 4 of 5 attempts. The researchers predict models under 30B could achieve end-to-end replication by the end of 2026.
Realtime-Bench: Frontier LLMs achieve only 44% success on real-time interactive tasks
Researchers from NVIDIA, UT-Austin, and other institutions released Realtime-Bench, a benchmark of 50 interactive game and web tasks that require real-time decision-making. Testing 12 frontier models, the best performer (GPT-6-Astra) achieved only 44% success, highlighting LLMs' weakness in low-latency tasks.
Thinking Machines Lab debuts Inkling 2 Small, 594B-param MoE; weights due next week
Thinking Machines Lab introduced Inkling 2 Small, a mixture-of-experts model with 594B total and 17B active parameters, initially available as a proprietary model via its Tinker platform and Tinker Playground. The company plans to release the weights next week after safety testing. Artificial Analysis reports the model matches Gemini 3.8 Flash (high) in its index and would be the top open model outside China once weights are released.
2 reports · 1 sources
News updates every hour. Subscribe in your RSS reader to get new stories automatically.
Subscribe