Qwen 3.5 35B A3B (Non-reasoning): benchmark results
Qwen 3.5 35B A3B evaluated with reasoning disabled. Provider: Alibaba. Released 2026-02-24. Access: Open.
Aggregate Rating 1557 ± 9, rank #842 of 2168 rated models, from 230 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| UAV-DualCog - Landmark-Driven Action Decision | 57.5 | Answer accuracy (%; 1,024 questions asking which way the UAV | 94.3 |
| DistortBench - Blur | 66.6 | Joint accuracy (%) on the three blur types (Gaussian, lens, | 94.1 |
| PerSo - Behavioral | 75 | Accuracy (%) | 92.9 |
| Chronicles-OCR - Mature Text Parsing | 0.71 | Paragraph-level normalized edit similarity (0-1: one minus t | 89.3 |
| PolicyShiftBench - Adaptive Split | 63.7 | F1 of the block decision (%; policy-conditioned image guardr | 88.9 |
| DistortBench | 56.8 | Joint accuracy (%) on DistortBench's 13,500 four-choice ques | 88.2 |
| DistortBench - Luminance | 51.5 | Joint accuracy (%) on the four luminance distortions (bright | 88.2 |
| DistortBench - Noise | 60.5 | Joint accuracy (%) on the five noise distortions (white, col | 88.2 |
| CPTU Bench | 4.18 | Average Score (1-5) | 87.1 |
| Chronicles-OCR - Archaic Text Parsing | 0.2 | Paragraph-level normalized edit similarity (0-1: one minus t | 85.7 |
| Video-HOCA - Physical Reasoning (Ontological) | 75.8 | Rubric score (0-100) by a GPT-OSS-120B judge of the structur | 84.2 |
| Video-HOCA - Plausibility Check | 82.9 | Accuracy (%) of the binary plausibility judgement on Video-H | 84.2 |
Interactive version: theaggregate.ai/model?slug=qwen-3-5-35b-a3b-non-reasoning · How It Works · Data refreshed daily, snapshot 2026-10-20.