Qwen3-235B-A22B-Instruct-2507-tput: benchmark results
Provider: Alibaba.
Aggregate Rating 1661 ± 40, rank #550 of 10916 rated models, from 17 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| HalluScore | 0.73 | Faith% (self-reported) | 96.9 |
| DSGym - DSBio | 38.89 | Exact-Match Accuracy (%) | 70.8 |
| HalluScore - Adversarial | 53.61 | Factual hallucination rate (%) on the 360 adversarial HalluS | 68.8 |
| DSGym - DSPredict Hard - Any Medal | 2.4 | Any Medal Rate (%; private leaderboard) | 63.6 |
| DSGym - DABStep Hard | 17.46 | Exact-Match Accuracy (%) | 58.3 |
| HalluScore - Factual Hallucination | 59.49 | Factual hallucination rate (%): share of the 827 HalluScore | 43.8 |
| HalluScore - History | 67.08 | Factual hallucination rate (%) on the 240 historical HalluSc | 43.8 |
| DSGym - DABStep Easy | 73.61 | Exact-Match Accuracy (%) | 41.7 |
| HalluScore - Reasoning | 61.36 | Factual hallucination rate (%) on the 220 HalluScore questio | 37.5 |
| DSGym - DAEval-Verified | 85.08 | Exact-Match Accuracy (%) | 33.3 |
| DSGym - DSPredict Easy - Above Median | 33.3 | Above Median Rate (%; private leaderboard) | 31.8 |
| DSGym - DSPredict Easy - Percentile | 42.9 | Mean Leaderboard Percentile (%; private leaderboard) | 31.8 |
Interactive version: theaggregate.ai/model?slug=qwen3-235b-a22b-instruct-2507-tput · How It Works · Data refreshed daily, snapshot 2026-11-08.