The open model release just went weights-first. shanghai ai lab put atria dawn preview out on…
the open model release just went weights-first. shanghai ai lab put atria dawn preview out on september 11 under the mit license, checkpoint and code on github before any wrapper or api.
the release channel is the repo now.
Context
The Hugging Face model card for Atria Dawn Preview, published by Shanghai Artificial Intelligence Laboratory under the internlm organization, says it is built on the 744B-parameter MoE GLM-5.2 foundation model, with 256K context, text input only, and BF16 and FP8 variants. Its license section says the code and model weights in the repository are released under the MIT License. An arXiv paper titled Atria Dawn: The Dawn of Agentic Superintelligence introduces the model as a foundation agentic language model. A secondary report, PacketNebula, dated September 15, 2026, gives September 11 as the repository creation date.
The weights and an MIT card are supported, and they cover the Hugging Face repository contents. Training data and full training recipe availability were not read, so open weights are not shown to be fully open source, and the model is a derivative of another lab's base. The September 11 date comes from a secondary source and the model card could not be dated. Checkpoint and code on github is not supported as worded, since the weights are on Hugging Face and no GitHub URL was observed. The card lists hosted API access, and whether an API came after the weights is not established. Benchmark rows are vendor-reported. The release channel is the repo now is the author's line.
Related work
- Shanghai ai lab released atria dawn preview on september 11 ↗The earlier note on the same model release.
Watch next
- The GitHub repository and the training recipe.
Sources
- Hugging Face: internlm/Atria-Dawn-Previewhuggingface.co
- arXiv: Atria Dawnarxiv.org
- PacketNebula: Atria Dawn Preview (September 15, 2026)packetnebula.com
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 22 September 2026 at 07:52 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 →