Student-heavy AtriaASI debuts Atria Dawn, a base model trained on verified real-world experience
xiaohu · x · 2026-09-15
Blogger xiaohu spotlights AtriaASI's "Atria Dawn Preview," a base model for scientific research and complex agent tasks. About two-thirds of its 96 core members are students from Fudan, CAS, ECNU, HIT and other institutions, working with the Shanghai AI Lab.
Key claim: a "verifiable experience pipeline" — only real-world-tested, metric-validated, evidence-closed experience gets distilled into the model, which can also learn from failures. Claimed demos: a mini OS (MiniOS) built in 20 minutes, rule-driven playable games, 3D reconstruction of the Grand View Garden from Dream of the Red Chamber, natural-language 3D CAD (four-cylinder engine, humanoid robot, Long March 7 rocket) with tolerance and simulation, and 1-minute offline global week-ahead weather forecasting. Ships with a dedicated agent harness; 100M free tokens on signup. Unverified marketing claims.
More from coding & agent
- AI agent finds $3,500/year car insurance savings in 5 minutes, buys it and cancels old policy — garrytan · 2026-09-15
- Pokee Isaac MCP now runs inside OpenCode, generating complete artifacts in one pass — Kyrannio · 2026-09-15
- DB veteran Craig Mullins: recovery beats backup, and test envs must match production scale — craigmullins · 2026-09-15
- Craig Mullins: the database remembers every bad design decision, for decades — craigmullins · 2026-09-15
- Craig Mullins: indexes aren't free, and the optimizer is only as smart as your statistics — craigmullins · 2026-09-15
- Craig Mullins: database performance is an ongoing discipline, not a one-time project — craigmullins · 2026-09-15