Qwen 3 30B A3B: benchmark results
Can switch between direct answers and explicit reasoning within the same conversation, without loading another checkpoint. Training covers multilingual dialogue, code and tool use. Provider: Alibaba. Released 2025-04-29. Access: Open.
Aggregate Rating 1585 ± 5, rank #493 of 1656 rated models, from 988 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AgentEHR (ReAct) - Transfers | 27.72 | F1 (%; predicted transfers label set against the next record | 100 |
| DNBench - BIRD - Real World - Zero-shot | 22.4 | DNB-Score (0-100) | 100 |
| DNBench - BIRD - Single - Zero-shot | 47.4 | DNB-Score (0-100) | 100 |
| DNBench - Component - Violation F1 | 67.8 | Component Score (0-100) | 100 |
| DNBench - Real World - Zero-shot | 25.3 | DNB-Score (0-100) | 100 |
| DNBench - Single - Zero-shot | 48.9 | DNB-Score (0-100) | 100 |
| DNBench - Spider - Real World - Zero-shot | 26.5 | DNB-Score (0-100) | 100 |
| DNBench - Spider - Single - Zero-shot | 49.6 | DNB-Score (0-100) | 100 |
| PersonalHomeBench - Counterfactual Reasoning (Sole Reasoning) | 74.53 | Accuracy (%) on 2,000 counterfactual questions, macro-averag | 100 |
| PersonalHomeBench - Multiple Choice (Sole Reasoning) | 91.6 | Accuracy (%) on 1,000 multiple-choice household questions (e | 100 |
| STQA - Vanilla - Historical Query Accuracy | 64.06 | Accuracy (%) | 100 |
| STQA - Vanilla - Historical Query F1 | 67.55 | F1 (%) | 100 |
Interactive version: theaggregate.ai/model?slug=qwen-3-30b-a3b · How It Works · Data refreshed daily, snapshot 2026-10-20.