Atria Dawn Preview
Category scores
Scores per capability area with the model's rank among ranked models in that category.
- 1Agentic & tool use82.3
- 2Coding48.7
Benchmark results (11)
As reported in system cards, papers and public evaluations. Position compares this model with every other model reporting the same benchmark.
Agentic & tool use
| Terminal-Bench 2.1 | 78.3 | #32 of 68 |
| BrowseComp | 92.5 | #1 of 50 |
| JobBench | 50.3 | #14 of 34 |
| AutomationBench | 53.8 | #3 of 25 |
| CyberGym | 86.5 | #5 of 25 |
| τ³-bench results | 41.2 | #16 of 23 |
| DeepSearchQA | 96.0 | #1 of 20 |
| BFCL v4 | 77.0 | #1 of 19 |
| WideResearch | 81.9 | #3 of 16 |
| MLE-Bench Lite | 86.2 | #1 of 2 |
Coding
| SWE-bench Pro | 59.6 | #27 of 69 |
Pricing by source
USD per 1M tokens.
| Source | Input | Output | Cached in |
|---|---|---|---|
| List price | — | — | — |
Shanghai Artificial Intelligence Laboratory publishes the full-precision and FP8 Atria Dawn Preview checkpoints under MIT for self-hosting. The pricing catalog represents the open-weight row as self-host/free-per-token before infrastructure costs. Atria also offers the exact `Atria-Dawn-Preview` model through its hosted API, but its public site and API documentation do not publish a first-party token rate.
Speed
Measured by us through OpenRouter (median of trials).
Not available on OpenRouter, so we cannot measure it yet.