Qwen Plus: benchmark results
Alibaba Cloud's Qwen Plus, an API-only model in the Qwen family positioned between Qwen-Max and Qwen-Turbo for balanced performance and cost, supporting up to 1M tokens of context. Provider: Alibaba. Released 2025-01-25. Access: API.
Aggregate Rating 1619 ± 14, rank #393 of 1635 rated models, from 75 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Plan-RewardBench | 69.96 | Pairwise accuracy (%), macro-average of the seven scenario c | 100 |
| Plan-RewardBench - Multi-Turn Planning (Hard) | 68.77 | Pairwise accuracy (%) on the 73 hard multi-turn planning pai | 100 |
| Plan-RewardBench - Single-Turn Planning (Easy) | 84.55 | Pairwise accuracy (%) on the 144 easy single-turn planning p | 100 |
| TimeLitmus - Finance Event-Side Counterfactual Pairs | 61.3 | Counterfactual pair correctness (%; both independently evalu | 100 |
| TimeLitmus - Traffic Natural Prediction | 42.1 | Natural accuracy (%; 420 Traffic records of TraffiDent roadw | 100 |
| Plan-RewardBench - Single-Turn Planning (Hard) | 74.68 | Pairwise accuracy (%) on the 158 hard single-turn planning p | 92.5 |
| Plan-RewardBench - Error Recovery | 73.75 | Pairwise accuracy (%) on the 361 robust error-recovery pairs | 90 |
| Software Engineering Arena - Model Arena | 1002 | Elo Rating | 86.4 |
| ConflictBench (Text-Only) | 59.71 | Task success rate (%): share of scenarios in which the agent | 85.7 |
| ToolRobustBench - User Intent | 0.8 | Success rate (0-1, native; indirect, underspecified, distrac | 83.3 |
| Tinybird AI SQL Benchmark - First-Attempt Success Rate | 98 | Questions answered with a valid query on the first attempt ( | 80.9 |
| Phare - Bias Resistance | 55.7 | Score (%) | 80.3 |
Interactive version: theaggregate.ai/model?slug=qwen-plus · How It Works · Data refreshed daily, snapshot 2026-11-05.