← Models
Qwen3.8 Max
Alibaba Cloud / Qwen Team · open weight · qwen3.8-max
Qwen3.8 Max, released as the Qwen3.8-2.4T-A95B open-weight checkpoint, is Qwen's flagship mixture-of-experts model for coding, research, professional workflows, and long-horizon agents. It has 2.4 trillion total parameters with 95 billion active parameters, a native 262,144-token context window extensible to about 1 million tokens, and always-on thinking. QwenCloud serves the same model in the Qwen3.8 Max API family with text and image input.
Benchmark scores
| Benchmark | Score |
|---|---|
| QwenReactBench | 1724.00 |
| QwenSVG | 1713.00 |
| PaperBench | 0.93 |
| MRCR v2 (8-needle) | 0.93 |
| GPQA | 0.93 |
| VideoMME w sub. | 0.90 |
| RealWorldQA | 0.88 |
| Terminal-Bench 2.1 | 0.87 |
| OSWorld-Verified | 0.86 |
| AndroidWorld | 0.85 |
| MMMU-Pro | 0.82 |
| WideSearch | 0.82 |
| QwenSWEBench | 0.81 |
| MobileWorld | 0.78 |
| AndroidBench | 0.75 |
| CoWorkBench | 0.75 |
| FrontierSWE | 0.73 |
| Toolathlon | 0.72 |
| SkillsBench | 0.70 |
| Humanity's Last Exam | 0.44 |
Pricing
- DeepInfra$1.65 / $4.95
- Novita$2.00 / $6.00
- Fireworks$2.00 / $6.00
- Together$2.50 / $6.25
Input / output per 1M tokens
AA metrics
No Artificial Analysis link yet.
Arena Elo
- vision1302
- code1671
- text1481
