Shanghai ai lab released atria dawn preview on september 11, a 744 billion parameter agentic…
shanghai ai lab released atria dawn preview on september 11, a 744 billion parameter agentic mixture of experts model with mit weights, no paper, no blog post, just a github repo and a free api. at 1.5 terabytes of bf16 weights and a 1 million token window it is the largest permissively licensed model to date.
the loudest releases are now the quiet ones.
Context
The Hugging Face card for Atria Dawn Preview describes a preview release of a new-generation agentic model from the Shanghai Artificial Intelligence Laboratory, built on the 744B-parameter mixture-of-experts GLM-5.2 foundation model. Its model table lists 256K context for the Instruct model, with an FP8 variant, and a hosted API console exists. PacketNebula of 15 September 2026, a secondary source, reports the repository was created on 11 September 2026, that BF16 weights are about 1.5 TB and FP8 about 756 GB, and that the weights are MIT licensed.
The first-party card shows a 256K context window, not 1 million, so the note's window is not supported. The model is built on a GLM-5.2 base from another lab, not a from-scratch 744B model. PacketNebula also reports an arXiv paper on 14 September, which was not read, so no paper at release is not claimed here. The MIT license comes from the secondary source and the card's license line was not located. Largest permissively licensed model to date has no source read. Repository creation is not proof of public release. The benchmark table was an image and is vendor-reported. The loudest releases being the quiet ones is the author's opinion.
Watch next
- The arXiv paper, the license file and independent benchmarks.
Sources
- Atria-Dawn-Preview (Hugging Face)huggingface.co
- Atria Dawn Preview 744B MIT (PacketNebula, 15 Sep 2026)packetnebula.com
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 19 September 2026 at 01:17 IST. Sources are the papers and datasets the note draws on.
View the original post ↗Embed this note
More notes
The air is now being asked to keep its own ledger
the air is now being asked to keep its own ledger: ecmwf’s aifs compo becomes the first ai model to forecast atmospheric composition globally every three hours, cleanair simulates 365 days of pm2.5 over china in ten seconds, and a unified framework maps six pollutants at one kilometer across the whole country. the air now files its own composition report.
read the note →The current is now being asked to draw its own map
the current is now being asked to draw its own map: china’s langya 2.0 predicts six ocean phenomena including internal waves and mesoscale eddies, a deep net called wenhai resolves eddies globally with air sea flux formulas built in, and scripps infers surface currents from the way temperature patterns deform in satellite images. the ocean now files its own circulation report.
read the note →The soil is now being asked to report its own carbon
the soil is now being asked to report its own carbon: a nix color sensor paired with generative data augmentation predicts soil organic carbon without a lab, random forest drives 74 percent of soil health mapping studies, and sentinel 2 tracks five year carbon change across france and italy from 922 samples. the dirt now files its own carbon account.
read the note →