- Terminal Bench 2.0-Terminus (avg of 5 runs) 70.3
- SWE-Verified (internal scaffold) 77.7
- SWE-Pro (Qwen-corrected tasks) 57.6
- SWE-Multilingual 75.8
- NL2repo 41.1
- SciCode 51.3
- QwenWebDev 1536
- QwenSVG 1588
- Qwenclaw 61.8
- CoWorkBench 65.1
- ClawEval 62.7
- Skillsbench (OpenCode, 78 tasks, avg of 5) 54.9
- BFCL-V4 72.9
- MCP-Mark 58.7
- MCP-Atlas (public set) 73.2
- Vitabench (claude-4.5-sonnet judge) 45.6
- Deep-Planning 62.3
- SpreadSheetBench-v1 86.3
- Kernel Bench L3 2.06/98% (median speedup / share faster than torch.compile) 2.06
- QwenWorldBench 62.1
- GPQA Diamond 90.3
- HLE 34.7
- LiveCodeBench 89.6
- HMMT 2026 Feb 92.9
- IMOAnswerBench 86
- CritPT 6
- Apex 22.7
- MMLU-Pro 88.5
- MMLU-Redux 94.5
- SuperGPQA 71.4
- IFEval 94.6
- IFBench 79.1
- MRCR-v2 128k 91.7
- WMT24++ (XCOMET-XXL, 55 languages) 84.6
- MAXIFE 88.8
- MMMLU 89
- MMLU-ProX (29 languages) 85.4
- NOVA-63 58.8
- INCLUDE 83
- Global PIQA 90.3
- PolyMATH 84
- MMMU-Pro 79
- MathVision 90.3
- BabyVision 70.4 / 64.7 (with CI / without CI) 70.4
- CharXiv(RQ) 85.9 / 84.4 (with CI / without CI) 85.9
- HiPhO 84.1
- ERQA 69.8
- VisFactor 42.8
- MedXpertQA-MM 71
- ScreenSpot Pro (enable_thinking=False) 79
- OSWorld-Verified (enable_thinking=False) 73.3
- AndroidWorld 81
- QwenVision2Code 1772
- ClawEval-MM 55.7
- SimpleVQA (with search) 81.7
- WorldVQA (with search) 61.1
- MMSearchPlus (with search) 41.4
- BC-VL (with search) 51.1
- MMBC (with search) 46.3
- RealWorldQA 86.9
- CountQA 77
- OmniDocBench1.5 91.4
- OCR-Bench-V2(EN) 70.7, OCR-Bench-V2(ZH) 67.1 70.7
- ODinW13 51.1
- LingoQA 83.4
- Ego3D-Bench (lower is better) 5.9
- SURDS 77.2
- VLADBench 77.2
- VideoMME (w/ sub.) 88
- VideoMMMU 85.4
- MLVU (M-Avg) 87.4
- TVBench 78.2
- LVBench 76.2
Qwen3.7-Plus
Qwen3.7-Plus ranks first on 3 of the 92 benchmarks it has scores on.
Qwen Released 2 Jun 2026 106
Benchmarks with a score
3
Ranked first, out of five or more models
13
In the top three
Qwen Jun 2026 Qwen3.7-Plus
Closed weights
147.1
#74 of 271
Epoch Capabilities Index
- IFBench 79.1% #13 /70
- MMMU-Pro 79% #16 /72
- SciCode 51.3% #8 /47
Where it ranks
Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.
Knowledge and language
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| MMLU-Pro | 88.5% | #4 of 151 | Model card | |
| SimpleVQA | 81.7% | #1 of 23 | Model card | |
| IFEval | 94.6% | #6 of 126 | Model card | |
| MMLU-Redux | 94.5% | #4 of 38 | Model card | |
| MMLU-ProX | 85.4% | #2 of 17 | Model card | |
| SuperGPQA | 71.4% | #4 of 26 | Model card | |
| Phare | 74.1% | #10 of 62 | Official leaderboard | |
| LMArena (Elo) | 1,456 | #55 of 333 | Official leaderboard | |
| MRCR v2 | 91.7% | #5 of 27 | Model card | |
| IFBench | 79.1% | #13 of 70 | Model card | |
| MMMLU | 89% | #14 of 62 | Model card | |
| INCLUDE | 83% | #8 of 25 | Model card | |
| MAXIFE | 88.8% | #2 of 5 | Model card | |
| Humanity's Last Exam | 34.7% | #69 of 147 | Model card | |
| Extended NYT Connections | 74.8% | #47 of 97 | Official leaderboard | |
| NOVA-63 | 58.8% | #3 of 5 | Model card | |
| Global PIQA | 90.3% | #5 of 8 | Model card | |
| Document Arena | 1,453 | #25 of 37 | Official leaderboard |
Math and science
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| PolyMath | 84% | #2 of 18 | Model card | |
| GPQA Diamond | 87.9% | #56 of 219 | Epoch AI | |
| OTIS Mock AIME 2024-2025 | 93.3% | #52 of 202 | Epoch AI | |
| IMO-AnswerBench | 86% | #12 of 37 | Model card | |
| CritPt | 9.14% | #62 of 160 | Epoch AI | |
| HMMT 2026 | 92.9% | #7 of 18 | Model card | |
| DTBench | 73.3% | #72 of 178 | Official leaderboard | |
| HiPhO | 84.1% | #2 of 4 | Model card | |
| FrontierMath Tiers 1-3 v2 | 34.4% | #66 of 89 | Epoch AI | |
| MathArena (Apex, IMO, USAMO) | 22.7% | #12 of 13 | Model card |
Coding
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| LiveCodeBench | 89.6% | #3 of 73 | Model card | |
| SciCode | 51.3% | #8 of 47 | Model card | |
| SWE-bench Verified | 77.7% | #30 of 150 | Model card | |
| StructEval | 76.8% | #9 of 28 | Official leaderboard | |
| SWE-bench Multilingual | 75.8% | #28 of 70 | Model card | |
| SWE-Bench Pro | 57.6% | #34 of 77 | Model card | |
| Kernel Bench L3 | 2.06 | #1 of 2 | Model card | |
| Vibe Code Bench | 46.4% | #66 of 115 | Official leaderboard | |
| Monthly-SWEBench | 47% | #8 of 12 | Official leaderboard | |
| NL2Repo-Bench | 41.1% | #21 of 28 | Model card | |
| AndroidBench | 57.7% | #25 of 32 | Official leaderboard | |
| Code Migration | 12.9% | #62 of 78 | Official leaderboard | |
| FrontierCode | 10.2% | #44 of 46 | Epoch AI | |
| QwenSVG | 1,588 | #2 of 2 | Model card | |
| QwenWebDev | 1,536 | #2 of 2 | Model card | |
| QwenVision2Code | 1,772 | #1 of 1 | Model card |
Agents and tools
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Terminal-Bench 2.0 | 70.3% | #4 of 61 | Model card | |
| BFCL V4 | 72.9% | #4 of 24 | Model card | |
| DeepPlanning | 62.3% | #1 of 6 | Model card | |
| AndroidWorld | 81% | #5 of 18 | Model card | |
| Pencil Puzzle Bench | 16.7% | #15 of 53 | Official leaderboard | |
| OSWorld-Verified | 73.3% | #10 of 33 | Official leaderboard | |
| ScreenSpot-Pro | 79% | #8 of 25 | Model card | |
| VITA-Bench | 45.6% | #4 of 12 | Model card | |
| QwenClawBench | 61.8% | #2 of 6 | Model card | |
| MCPMark | 58.7% | #4 of 10 | Model card | |
| SnakeBench | 23 | #98 of 183 | Official leaderboard | |
| MCP Atlas | 73.2% | #24 of 40 | Model card | |
| SkillsBench | 54.9% | #8 of 13 | Model card | |
| Mystery Game Puzzles | 17% | #58 of 80 | Epoch AI | |
| Claw-Eval | 62.7% | #11 of 14 | Model card | |
| MM-BrowseComp | 46.3% | #4 of 5 | Model card | |
| Agent Arena | -0.07 | #43 of 53 | Official leaderboard | |
| RecreationBench | 9.15% | #10 of 10 | Official leaderboard | |
| OSWorld 2.0 | 2.80% | #7 of 7 | Official leaderboard | |
| BrowseComp-VL | 51.1% | #2 of 2 | Model card | |
| MMSearch-Plus | 41.4% | #1 of 1 | Model card |
Professional work
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| MedXpertQA-MM | 71% | #4 of 12 | Model card | |
| MortgageTax | 66.2% | #33 of 86 | Official leaderboard | |
| RadLE 2.0 | 472.0 | #6 of 12 | Official leaderboard | |
| SpreadsheetBench | 86.3% | #3 of 6 | Model card | |
| Excel Modeling Benchmark | 49.3% | #56 of 76 | Official leaderboard | |
| Legal Research Bench | 16.4% | #66 of 88 | Official leaderboard | |
| Tax Agent Bench | 10.5% | #64 of 79 | Official leaderboard | |
| Vals Finance Agent v2 | 38.2% | #65 of 80 | Official leaderboard | |
| Harvey LAB (Legal Agent Benchmark) | 0% | #76 of 80 | Official leaderboard | |
| CoWorkBench | 65.1% | #6 of 6 | Model card |
Vision, audio and robotics
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| OCRBench v2 | 70.7% | #2 of 20 | Model card | |
| WorldVQA | 61.1% | #1 of 9 | Model card | |
| MLVU | 87.4% | #2 of 16 | Model card | |
| RealWorldQA | 86.9% | #4 of 30 | Model card | |
| Video-MME | 88% | #6 of 42 | Model card | |
| MATH-Vision (MathVision) | 90.3% | #7 of 42 | Model card | |
| MMMU-Pro | 79% | #16 of 72 | Model card | |
| ODinW-13 | 51.1% | #2 of 9 | Model card | |
| Video-MMMU | 85.4% | #7 of 30 | Model card | |
| CharXiv | 85.9% | #17 of 67 | Model card | |
| Vision Arena | 1,263 | #35 of 136 | Official leaderboard | |
| ERQA | 69.8% | #7 of 26 | Model card | |
| BabyVision | 70.4% | #9 of 27 | Official leaderboard | |
| LVBench | 76.2% | #11 of 25 | Model card | |
| LingoQA | 83.4% | #2 of 4 | Model card | |
| QwenWorldBench | 62.1% | #1 of 2 | Model card | |
| TVBench | 78.2% | #4 of 7 | Model card | |
| CountQA | 77% | #2 of 3 | Model card | |
| Claw-Eval-MM | 55.7% | #5 of 7 | Model card | |
| VisFactor | 42.8% | #3 of 4 | Model card | |
| SURDS | 77.2% | #2 of 2 | Model card | |
| Ego3D-Bench | 5.90 | #1 of 1 | Model card | |
| VLADBench | 77.2% | #1 of 1 | Model card |
Safety and security
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Dictatorship Eval | 45.6% | #14 of 20 | Official leaderboard | |
| CyberBench (Vals) | 37.5% | #25 of 26 | Official leaderboard |
Other
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| SAGE v1.1 (Vals) | 89.2% | #6 of 28 | Official leaderboard | |
| LMCA | 37.6% | #60 of 139 | Epoch AI | |
| Conceptual Reasoning Index | 49.4 | #70 of 149 | Official leaderboard | |
| ACCoRD | 60.7 | #87 of 150 | Official leaderboard | |
| Vals Multimodal Index | 53.9% | #22 of 33 | Official leaderboard | |
| SAGE (Vals) | 39.3% | #59 of 81 | Official leaderboard |
Qwen's models over time
Qwen's models in color, against every other model on the Epoch Capabilities Index, by release date.
Its launch post
The benchmarks the lab chose to headline, with the scores it gave.