Zhongguancun releases MatOS, an open-source AI-scientist OS for materials research
量子位 · wechat · 2026-11-17
Zhongguancun Academy and its AI research institute released MatOS, an agent operating system for materials research that organizes datasets, knowledge, tools, skills, roles, and workflows into evolvable scientific assets with stateful workflows. The asset library holds 3,135 datasets, 386 skills, 104 workflows, and 51 expert roles, integrating 48 materials software packages; code is open source with an arXiv report. On ResearchClawBench with the same GPT-5.4 base it scores 33.6% higher than CodexCLI (80.8% higher on materials tasks), and hits 97.96% full-problem success on MatTools. A DeepSeek-V4-Flash configuration costs $0.85 per task vs CodexCLI's $2.70 while scoring higher. Case studies include screening 97 solid electrolyte candidates from 30M+ structures and developing DuoCompNet with 3.18°C MAE, 13.6% better than the runner-up.
More from coding & agent
- Agent Substrate: open-source zero-trust runtime suspends idle agents, fits 10x more per hardware — Saboo_Shubham_ · 2026-11-17
- ChatGPT Wallet solves secure payment, not buying the right thing — krishnan · 2026-11-17
- Grok bot phones a store and secures a $1,511 refund for its user — elonmusk · 2026-11-17
- Practical tips to cut agent memory costs on OpenAI: cache TTL, Flex, fewer calls — gabrielchua · 2026-11-17
- 10 open-source repos to build an entire AI content factory solo — aftahi_ai · 2026-11-17
- What 4 calibration papers do (and don't) say about gating agent tool calls on model confidence — Xjjjjyn · 2026-11-17