Qwen 3.5 35B A3B (Thinking): benchmark results
Provider: Alibaba. Released 2026-02-24. Access: Open.
Aggregate Rating 1647 ± 9, rank #529 of 2170 rated models, from 130 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Medmarks - LongHealth Task 1 | 91.83 | Score (%) | 98.9 |
| ChartArena - Bar Chart (English) | 46.2 | mAP-high (%) on English bar charts, each model parses the ch | 97.7 |
| ChartArena - Box Plot (Chinese) | 50.6 | mAP-high (%) on Chinese box plots, each model parses the cha | 97.6 |
| Medmarks - LongHealth Task 2 | 90.75 | Score (%) | 97.2 |
| Swallow - Japanese MT-Bench - Math | 99.3 | Judge Score (normalized, %) | 97 |
| Medmarks - PubMedQA | 80.6 | Score (%) | 96.6 |
| Medmarks - MedQA | 93.72 | Score (%) | 95.5 |
| ChartArena - Box Plot (English) | 42.2 | mAP-high (%) on English box plots, each model parses the cha | 95.3 |
| ChartArena - Line Chart (English) | 60.3 | mAP-high (%) on English line charts, each model parses the c | 95.3 |
| ChartArena - Pie Chart (English) | 89.7 | mAP-high (%) on English pie charts, each model parses the ch | 95.3 |
| ChartArena - English | 54.1 | mAP-high (%) averaged over the eight chart types (bar, line, | 95.2 |
| ChartArena - Pie Chart (Chinese) | 88.4 | mAP-high (%) on Chinese pie charts, each model parses the ch | 95.2 |
Interactive version: theaggregate.ai/model?slug=qwen-3-5-35b-a3b-thinking · How It Works · Data refreshed daily, snapshot 2026-11-05.