Microsoft AI's text-to-speech model for keeping one voice across 23 languages, with voice cloning from short audio
Atria Dawn Preview
New· 4Open-weights agentic AI model from the Shanghai lab behind InternLM, built on GLM-5.2
About Atria Dawn Preview
Open weights for an agentic model built on Z.ai's GLM-5.2 became public on 11 September 2026, when the Shanghai Artificial Intelligence Laboratory released Atria Dawn Preview. The same lab's Hugging Face organisation also publishes the InternLM models. An FP8 checkpoint, which uses lower numerical precision, arrived the next day.
Foundation and licence
Atria Dawn Preview builds on GLM-5.2, which uses a mixture-of-experts design and has 744 billion parameters. Code and weights are released under the MIT licence, which allows broad reuse. The context window is 256K tokens. Hosted API access is mentioned on the model card next to the downloadable weights, though nothing about price is published.
Four groups of ability
The model is meant for jobs that call for constant tool use and tasks finished in several steps. Its model card groups the abilities into four areas:
- Discovery: looking things up and gathering evidence.
- Creation: constructing software and machine-learning systems.
- Delivery: drafting documents and reports.
- Cybersecurity: security-oriented tasks.
Reported results
The model card includes a benchmark table, reported by the developer and not independently reproduced. Scores listed are 96.0 for DeepSearchQA, 92.5 for BrowseComp, 59.6 for SWE-bench Pro, 77.0 for BFCL v4 and 86.5 for CyberGym. The table sets the software-engineering score beside a higher one for another leading model, so the card itself shows where the gap lies. Those numbers describe a preview release, and readers should treat them as the laboratory's claims.
Who might use it
Teams that want to run an agentic model on their own hardware, adapt it, or build tools around research, coding and report-writing workflows are the natural audience, given the open weights and permissive licence. The FP8 checkpoint offers a lighter option for those with limited memory.
TrustList summary
Atria Dawn Preview is an agentic AI model whose weights were made public on 11 September 2026. The Shanghai Artificial Intelligence Laboratory, also the publisher of the InternLM family, is behind it. It builds on Z.ai's GLM-5.2, a mixture-of-experts model with 744 billion parameters. It targets work that needs continuous tool use and multi-step task completion.
What Atria Dawn Preview offers
- Discovery: research and evidence retrieval.
- Creation: building software and machine-learning systems.
- Delivery: producing documents and reports.
- Cybersecurity: security-related tasks, one of the abilities named on its model card.
- Open release: code and weights are published under the MIT licence, and a lower-precision FP8 checkpoint followed a day later.
Key facts
- Released: 11 September 2026
- Developer: Shanghai Artificial Intelligence Laboratory
- Context window: 256K tokens
- Foundation model: GLM-5.2 from Z.ai
- Benchmarks: the model card's results are reported by the developer and have not been independently reproduced
Summary written by TrustList from the business's own published information, Nov 2026.
Benchmarks & AI stats
| Benchmark | Official | Community avg |
|---|---|---|
| CyberGymreasoning | 86.5 % | - |
“Official” values are editor-approved and feed the ranking. “Community avg” is the mean of member submissions (shown for transparency; it never affects the ranking until an editor approves a value).
Sign in to add a benchmark score for this model.
Atria Dawn Preview launches
Trust Score
An earned signal from verification, reviews, awards, transparency and engagement - the vendor can't buy it.
Updated 9/16/2026
What the business tells TrustList
- Founded
- 2026 (0 years) evidence sent
Request a demo or quote from Atria Dawn Preview
Reviews
Write the first review of Atria Dawn Preview
Used it? Your experience helps other buyers decide.
Write a reviewQuestions & answers
No questions yet. Be the first to ask about Atria Dawn Preview.
Atria Dawn Preview alternatives
Similar listings buyers compare against this one.
Anthropic's smallest and fastest Claude 5.5 model, available on its own API and three cloud platforms
Liquid AI's open-weight decision model, giving yes or no, choice or score answers from text and images
Aleph Alpha's open-weight English-German mixture-of-experts model for self-hosting, released under Apache 2.0 in October 2026
Microsoft AI's real-time speech-to-text model for voice agents, transcribing 60 languages with continuous language detection
Perplexity's open-weight decision model that picks among given options for routing and checks in AI systems
Cantina Security's first open-weight security model, a post-trained GLM-5.3-Flash for vulnerability research and exploit checks
Google's open multimodal embedding model that maps text, images, audio and video into one vector space, under Apache 2.0
More on TrustList
Everything here links back to the same verified catalogue. Pick your next stop.
- Write a reviewHelp other buyers decide
- 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.
- LaunchesNew products and releases, voted up by the community every day.
- 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.