Benchmark
MMStar
Vision benchmarks measure how well models understand images. Scores are percentages. Higher is better.
- Models measured
- 68
- Scores recorded
- 72
- Sources
- 3
- IQ weight
- 0.99
Score distribution
- 65.1%
- Mean
- 67.1%
- Median
- 13.1%
- Std dev
- 47.5%
- 10th percentile
- 81.4%
- 90th percentile
- 24.9%
- Min
This benchmark has a weight of 0.99 for chat models, measured over 67 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model | Score |
|---|---|---|
| 1 | Qwen3.8 MaxQwen3.8-Max release vision/video table · published setting· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 85.9% |
| 2 | Qwen3.5 397B A17BQwen3.5 release/model card · published setting· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 83.8% |
| 3 | Qwen3.6 Plus· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 83.3% |
| 4 | Qwen3.6 PlusQwen official release table · thinking mode unless benchmark requires otherwise· Reported by Alibaba Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 83.3% |
| 5 | Qwen3.5 122B-A10B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 82.9% |
| 6 | Qwen3.5 35B A3B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 81.9% |
| 7 | Qwen3.6 27B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 81.4% |
| 8 | Qwen3.6 27BQwen official release table · thinking mode unless benchmark requires otherwise· Reported by Alibaba Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 81.4% |
| 9 | Qwen3.5 27B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 81% |
| 10 | Doubao Seed 1.8· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 79.9% |
| 11 | Qwen3 VL 32B Thinking· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. | 79.4% |
| 12 | Qwen3.5 Omni PlusQwen3.5-Omni release · published setting· Reported by Qwen, Qwen Team· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 79.4% |
| 13 | Seed 2.0 Mini· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 79.1% |
| 14 | Qwen3 VL 235Bthinking· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. | 78.7% |
| 15 | Qwen3 VL 235B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. | 78.4% |
| 16 | Qwen3 VL 32B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. | 77.7% |
| 17 | GPT 5· OpenVLM Leaderboard data | 75.7% |
| 18 | Qwen3.5 Omni FlashQwen3.5-Omni release · published setting· Reported by Qwen, Qwen Team· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 75.7% |
| 19 | qwen3-vl-30bthinking· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. | 75.5% |
| 20 | Qwen: Qwen3 VL 8B Thinking· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. | 75.3% |
| 21 | SenseNova-V6-Pro· OpenVLM Leaderboard data | 73.7% |
| 22 | Gemini 2.5 Pro· OpenVLM Leaderboard data | 73.6% |
| 23 | InternVL3 78B· OpenVLM Leaderboard data | 73.4% |
| 24 | GPT-5 mini· OpenVLM Leaderboard data | 73.2% |
| 25 | Hunyuan Standard Vision· OpenVLM Leaderboard data | 72.9% |
| 26 | internvl3-38b· OpenVLM Leaderboard data | 72.6% |
| 27 | qwen3-vl-30b· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. | 72.1% |
| 28 | Qwen3 VL 8B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. | 70.9% |
| 29 | Qwen2.5 VL 72B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claimValues from 2 sources
| 70.8% |
| 30 | ChatGPT-4o Latest· OpenVLM Leaderboard data | 70.2% |
| 31 | GPT 4.1· OpenVLM Leaderboard data | 69.8% |
| 32 | Gemini 2.0 Flash· OpenVLM Leaderboard data | 69.4% |
| 33 | GPT-4.5 Preview· OpenVLM Leaderboard data | 69.3% |
| 34 | internvl3-8b· OpenVLM Leaderboard data | 68.7% |
| 35 | Qwen2-VL-72B· OpenVLM Leaderboard data | 68.6% |
| 36 | GPT-5 nano· OpenVLM Leaderboard data | 68.3% |
| 37 | gemini-1.5-pro-002· OpenVLM Leaderboard data | 67.1% |
| 38 | GLM 4 Vision Plus· OpenVLM Leaderboard data | 66.3% |
| 39 | Grok 2 Vision 1212· OpenVLM Leaderboard data | 65.2% |
| 40 | gemini-1.5-flash-002· OpenVLM Leaderboard data | 64.4% |
| 41 | Qwen2.5-VL 7B Instruct· OpenVLM Leaderboard dataValues from 2 sources
| 64.1% |
| 42 | Qwen2.5-Omni 7B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 64% |
| 43 | Molmo 72B· OpenVLM Leaderboard data | 63.3% |
| 44 | DeepSeek VL2· OpenVLM Leaderboard dataValues from 2 sources
| 61.9% |
| 45 | InternVL2 26B· OpenVLM Leaderboard data | 61% |
| 46 | GPT-4.1 mini· OpenVLM Leaderboard data | 60.9% |
| 47 | Yi Vision· OpenVLM Leaderboard data | 60.5% |
| 48 | Moonshot V1 8K· OpenVLM Leaderboard data | 60.1% |
| 49 | Gemma 3 27B IT· OpenVLM Leaderboard data | 59.6% |
| 50 | Gemini 1.5 Pro· OpenVLM Leaderboard data | 59.1% |
Showing 1–50 of 72 results.
Data credits: OpenVLM Leaderboard (OpenCompass / VLMEvalKit), Apache-2.0 (licence) · Data by LLM Stats · BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links.