Shanghai AI Laboratory model product
Atria Dawn is a preview-stage agentic model from Shanghai AI Laboratory with a 744B-parameter MoE architecture and a 256K-token context window.
Updated Oct 23, 2026. Default version: Atria Dawn Preview
This profile uses the model's current scored version. Arena ratings and prices are shown separately.
Benchmark scores for Atria Dawn Preview.
Benchmark | Score | Rank | Participants | Percentile | Evidence | Evaluated |
|---|
| BenchmarkBFCL-V4 | Score77.00% | Rank01 | Participants22 | Percentile100.00% | EvidenceC | Evaluated |
| BenchmarkBrowseComp | Score92.50% | Rank01 | Participants70 | Percentile100.00% | EvidenceC | Evaluated |
| BenchmarkDeepSearchQA | Score96.00% | Rank01 | Participants13 | Percentile100.00% | EvidenceC | Evaluated |
| BenchmarkMLE-Bench Lite | Score86.20% | Rank01 | Participants2 | Percentile100.00% | EvidenceC | Evaluated |
| BenchmarkTau3 Banking | Score41.20% | Rank01 | Participants16 | Percentile100.00% | EvidenceC | Evaluated |
| BenchmarkSkillsBench | Score66.40% | Rank03 | Participants12 | Percentile81.82% | EvidenceC | Evaluated |
| BenchmarkWideSearch | Score81.90% | Rank03 | Participants15 | Percentile85.71% | EvidenceC | Evaluated |
| BenchmarkAutomationBench | Score53.80% | Rank04 | Participants27 | Percentile88.46% | EvidenceC | Evaluated |
| BenchmarkCyberGym | Score86.50% | Rank05 | Participants23 | Percentile81.82% | EvidenceC | Evaluated |
| BenchmarkGDPval-AA | Score1,583.00 points | Rank07 | Participants17 | Percentile62.50% | EvidenceC | Evaluated |
| BenchmarkSWE-Bench Pro | Score59.60% | Rank22 | Participants63 | Percentile66.13% | EvidenceC | Evaluated |
| BenchmarkTerminal-Bench 2.1 | Score78.30% | Rank25 | Participants48 | Percentile48.94% | EvidenceC | Evaluated |
Preference and agent-evaluation results for the default version.
The default version does not have a matching Arena result yet.
Official vendor API pricing appears first, followed by individual provider offers.
The default version has no current input or output token prices.
Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.
Provider-specific output speed and catalog latency for Atria Dawn Preview. Runtime does not affect the capability score.
No provider-specific speed or latency record is linked to the default version yet.
Output Speed is generated output tokens received per second. Catalog latency is reported separately from observed provider TTFT.
Technical details for the model's default version.
Available versions of this model. The score column identifies the version used in the overall ranking.
Version | Released | LLMBoard | Parameters | Context | Max output | Open weights | License |
|---|
| VersionAtria Dawn Preview | Released | LLMBoard83.28 | Parameters744B | ContextN/A | Max outputN/A | Open weightsYes | LicenseMIT |
Recommendations prioritize the same model type and family, then the closest LLMBoard score.
Key information about Atria Dawn and its available data.
Atria Dawn is a text-only Instruct model from Shanghai AI Laboratory for research and engineering workflows involving environment understanding, tool use, and multi-step task completion. Its weights are available in BF16 and FP8 formats under the MIT license.
Data as of 2026-10-22.
Common questions about Atria Dawn.
Atria Dawn's default version was released on Sep 11, 2026.
No official standard PAYG price is currently available for Atria Dawn.
Atria Dawn was created by Shanghai AI Laboratory.
A context window is not available for the default version.
Yes. The default version is marked as open weight under MIT.
No provider offering is currently linked to the default version.