Qwen3 Swallow 32B: benchmark results
Provider: Alibaba.
Aggregate Rating 1619 ± 37, rank #841 of 10916 rated models, from 4 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| FrameBench - Frame Identification - Japanese | 76.3 | Accuracy (%; Japanese FrameNet candidate frames) | 86 |
| FrameBench - Japanese | 93.2 | Accuracy (%; mean of five prompt templates) | 72 |
| Benchmarking Large Language Models for Graphem | 21.59 | Direct (self-reported) | 52.8 |
| Japanese G2P (Parse Mode) | 9.3 | Kana character error rate (%; can exceed 100) in parse mode | 47.2 |
Interactive version: theaggregate.ai/model?slug=qwen3-swallow-32b · How It Works · Data refreshed daily, snapshot 2026-11-08.