Qwen3.5-122B-A10B-FP8: benchmark results
Provider: Alibaba. Access: Open.
Aggregate Rating 1691 ± 29, rank #391 of 10916 rated models, from 34 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Video-HOCA - Plausibility Check | 88 | Accuracy (%) of the binary plausibility judgement on Video-H | 100 |
| Video-HOCA - Fine-grained Recognition (Ontological) | 85.9 | Macro-F1 (%) of choosing the correct anomaly description amo | 94.7 |
| 3DVerifier - Fine Geometry | 53.3 | Balanced accuracy (%) | 94.4 |
| Video-HOCA - Fine-grained Recognition (Causal) | 79.2 | Macro-F1 (%) of choosing the correct anomaly description amo | 89.5 |
| 3DVerifier - Material | 70.8 | Balanced accuracy (%) | 88.9 |
| ElecVQA-Bench - Binary Defect Assessment (Zero-Shot) | 71 | Balanced accuracy (%; mean of the defective and normal class | 83.3 |
| Video-HOCA - Category Attribution (Ontological) | 45 | Macro-F1 (%) of attributing each anomalous video to one of t | 78.9 |
| Trip+ - Soft Preferences | 63.19 | Soft-preference satisfaction (0-100), a rate scaled by 100: | 76.5 |
| 3DVerifier - Grasp | 58 | Balanced accuracy (%) | 66.7 |
| 3DVerifier - Part | 64.7 | Balanced accuracy (%) | 66.7 |
| 3DVerifier - Part Grounding | 75.7 | Balanced accuracy (%) | 66.7 |
| Video-HOCA - Physical Reasoning (Causal) | 61.4 | Rubric score (0-100) by a GPT-OSS-120B judge of the structur | 63.2 |
Interactive version: theaggregate.ai/model?slug=qwen3-5-122b-a10b-fp8 · How It Works · Data refreshed daily, snapshot 2026-11-08.