Atria Dawn Preview
An MIT-licensed 744B agentic model built on GLM-5.2, released as open weights by the lab behind InternLM.
About this launch
Atria Dawn Preview is an agentic model released as open weights on 11 September 2026 by the Shanghai Artificial Intelligence Laboratory, whose Hugging Face organisation also publishes the InternLM models, with an FP8 checkpoint following a day later. It is built on Z.ai's 744B-parameter mixture-of-experts GLM-5.2 foundation model and aimed at work that needs continuous tool use and multi-step task completion: the model card groups its abilities as discovery (research and evidence retrieval), creation (software and machine-learning systems), delivery (documents and reports) and cybersecurity. The context window is 256K tokens, and the code and weights are released under the MIT licence. The card lists hosted API access alongside the downloadable weights, but no price has been published. The card's benchmark table is vendor-reported and has not been independently reproduced. It gives 96.0 on DeepSearchQA, 92.5 on BrowseComp, 59.6 on SWE-bench Pro (against 74.7 for Claude Opus 5 in the same table), 77.0 on BFCL v4 and 86.5 on CyberGym. Only the CyberGym figure is stored as a score on this board, because it is the one row whose settings could be checked against a score already here; the others are recorded in this description only.
Discussion
Questions & answers about Atria Dawn Preview.
No questions yet. Be the first to ask about Atria Dawn Preview.
More on TrustList
Everything here links back to the same verified catalogue. Pick your next stop.
- All launchesToday's board, this week's and the all-time archive.
- Atria Dawn Preview on TrustListReviews, pricing, features and alternatives.
- CompaniesAgencies, consultancies and IT service providers, ranked by verified reviews.
- ProductsSoftware and SaaS with pricing, features, integrations and alternatives.
- ArticlesGuides, comparisons and research from the editorial desk and the community.
- AwardsAnnual recognition decided by verified reviews and an independent jury.
- AI ModelsBenchmark scores and community ratings for every major model.
- RequestsBuyers describe what they need; vendors respond directly.
- PeopleReviewers, authors and makers with public profiles.
- ComparePut up to four listings side by side before you shortlist.