Qwen 3.5 122B A10B (Non-reasoning): benchmark results
Qwen 3.5 122B A10B evaluated with reasoning disabled. Provider: Alibaba. Released 2026-02-24. Access: Open.
Aggregate Rating 1579 ± 11, rank #768 of 2168 rated models, from 149 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| UAV-DualCog - Future Observation Prediction | 37.8 | Answer accuracy (%; 1,024 questions asking which view the UA | 94.1 |
| EPIC-Bench - Embodied Compositional Attributes | 42.31 | Target localization score (0-100) on the embodied compositio | 93.2 |
| CPTU Bench | 4.26 | Average Score (1-5) | 91.1 |
| TableVista - Simulated Photo | 53.5 | Accuracy (%) on the 3,000 TableVista questions under the vis | 91.1 |
| EPIC-Bench - Ground Detection | 49.65 | Navigation score (0-100) on ground detection (localizing tra | 89.8 |
| TableVista - Screen Capture | 54.3 | Accuracy (%) on the 3,000 TableVista questions under the vis | 89.3 |
| Sci-Rho - Biology (English) | 67.8 | Average-case accuracy (%) on the English biology templates o | 87.5 |
| EPIC-Bench - Basic Attributes | 50.77 | Target localization score (0-100) on the basic-attribute tas | 84.1 |
| TableVista - Partial | 47.3 | Accuracy (%) on the 3,000 TableVista questions under the Par | 82.1 |
| TableVista - Structural Noise | 48.7 | Accuracy (%) on the 3,000 TableVista questions under the Str | 82.1 |
| EPIC-Bench | 44.54 | Average score (0-100) over all samples, each sample scored b | 81.8 |
| Sci-Rho | 80.8 | Average-case accuracy (%) over all seven languages (English, | 81.2 |
Interactive version: theaggregate.ai/model?slug=qwen-3-5-122b-a10b-non-reasoning · How It Works · Data refreshed daily, snapshot 2026-10-20.