Atria Dawn Preview
- Input
- —
- Output
- —
- Speed
- ~81tok/sest.
- Context
- 262Ktokens
Agentic model from Shanghai AI Lab built on GLM-5.2 (744B MoE). No API price found. Card also reports self-run Terminal-Bench 2.1, AutomationBench, GDPval and tau3-Banking, not mapped (our ids hold AA / Terminal-Bench 4.0 runs).
Rankings
Self-hostingEstimated
How we estimate →What it would cost to run these open weights yourself on rented GPUs. No API sells this model, so this is its price on the chart.
- Hardware
- 8× B200 180GB
- FP8 weights
- Throughput
- ~18,596 tok/s
- many requests batched
- Price
- $0.30–2.69 /M tok
- busy → light use
- On your own machine
- Mac Studio M5 Ultra 512GB (int4, ~42 tok/s single-stream)
- single consumer GPU or Mac
Assumptions (5)
- FP8 weights (744B params, 40B active per token) + 25% KV-cache headroom ≈ 930 GB
- 8× B200 180GB at $5.98–$14.24/GPU-hour on-demand (2026-09-24)
- ~18,596 output tok/s aggregate at batch 532 (bandwidth-bound); MoE compute scales with active params
- Blended 3:1 input:output; prefill ~212,625 tok/s
- Low = 75% utilization at the low GPU price; high = 20% at the high price
Benchmark scores
~ italic, dashed = no published score yet, estimated from related benchmarks.
Coding
SWE-bench VerifiedEstimated
~#6 of 21 (estimated)~80.2%
- #1Claude Opus 596.0%
- #2Claude Sonnet 585.2%
- #5MiniMax-M380.5%
- ~#6Atria Dawn PreviewEstimated80.2%
- #6Hy378.0%
No published score — estimated from related benchmarks
SWE-bench Pro (Public)Lab-reported
#12 of 2359.6%
- #1Claude Fable 5.293.0%
- #2Claude Opus 5.589.9%
- #11Nex-N2.5-Pro61.2%
- #12Atria Dawn Preview59.6%
- #13Laguna S 2.159.4%
huggingface.co · 2026-09-11
DeepSWE v1.1Estimated
~#38 of 46 (estimated)~47.2%
- #1Gemini 4 Argon78.8%
- #2DeepSeek V4.1 Flash 101777.0%
- #37Qwen3.8 Max (0902)51.0%
- ~#38Atria Dawn PreviewEstimated47.2%
- #38Ripple-65B Preview44.3%
No published score — estimated from related benchmarks
Terminal-Bench 4.0Estimated
~#29 of 102 (estimated)~22.7%
- #1Kimi K3.1 Pro67.2%
- #2Claude Sonnet 5.563.6%
- #28Qwen3.8 Flash-Next25.3%
- ~#29Atria Dawn PreviewEstimated22.7%
- #29MiMo-V2.6-Flash22.7%
No published score — estimated from related benchmarks
LiveCodeBenchEstimated
~#10 of 28 (estimated)~83.2%
- #1Qwen3.8-Omni-Flash92.6%
- #2Qwen3.8 Flash-Next91.9%
- #9Nemotron 3.5 Super VL NVFP487.4%
- ~#10Atria Dawn PreviewEstimated83.2%
- #10Ternary Bonsai 27B82.8%
No published score — estimated from related benchmarks
SciCodeEstimated
~#27 of 119 (estimated)~55.5%
- #1Claude Fable 5.267.7%
- #2Claude Opus 5.566.9%
- #26GPT-6.1 Sol55.7%
- ~#27Atria Dawn PreviewEstimated55.5%
- #27Claude Haiku 5.555.0%
No published score — estimated from related benchmarks
AA Coding Agent IndexEstimated
~#9 of 26 (estimated)~53.4
- #1Claude Opus 5.561.7
- #2Claude Opus 559.7
- #8GPT-6.1 Sol54.1
- ~#9Atria Dawn PreviewEstimated53.4
- #9Gemini 4 Argon53.2
No published score — estimated from related benchmarks
Agents
τ²-bench (Telecom)Estimated
~#38 of 120 (estimated)~76.0%
- #1Step 3.7 Flash98.5%
- #2DeepSeek-V4-Pro-0813-nvfp4-DSpark98.3%
- #37EXAONE 4.5 33B78.1%
- ~#38Atria Dawn PreviewEstimated76.0%
- #38GPT-5.4 nano76.0%
No published score — estimated from related benchmarks
OSWorld-VerifiedEstimated
~#3 of 15 (estimated)~83.3%
No published score — estimated from related benchmarks
OSWorld 2.0Estimated
~#3 of 5 (estimated)~57.9%
No published score — estimated from related benchmarks
GDPval-AA (Elo)Estimated
~#22 of 92 (estimated)~1456 Elo
- #1Claude Opus 5.51866
- #2Muse Wonder Full1857
- #21GPT-6 Sol1487
- ~#22Atria Dawn PreviewEstimated1456
- #22Claude Sonnet 51449
No published score — estimated from related benchmarks
AutomationBench-AAEstimated
~#29 of 85 (estimated)~38.6%
- #1Claude Fable 5.276.4%
- #2Muse Wonder Full71.3%
- #28GPT-5.547.3%
- ~#29Atria Dawn PreviewEstimated38.6%
- #29K2 Horizon 375B A23B37.2%
No published score — estimated from related benchmarks
τ-Bench Banking (AA)Estimated
~#22 of 91 (estimated)~31.3%
- #1Muse Spark 1.350.5%
- #2GLM-5.350.3%
- #21K2 Horizon MoVA 36B A4B32.0%
- ~#22Atria Dawn PreviewEstimated31.3%
- #22Inkling29.1%
No published score — estimated from related benchmarks