Benchmark
AI2D
Vision benchmarks measure how well models understand images. Scores are percentages.
- Models measured
- 69
- Scores recorded
- 71
- Sources
- 2
- IQ weight
- 1.11
Score distribution
- 82.4%
- Mean
- 84.2%
- Median
- 10%
- Std dev
- 70.7%
- 10th percentile
- 92.7%
- 90th percentile
- 25.6%
- Min
This benchmark has a weight of 1.11 for chat models, measured over 68 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 | Claude 3.5 Sonnet· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 94.7% |
| 2 | Qwen3.6 Plus· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 94.4% |
| 3 | GPT 4o· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 94.2% |
| 4 | Pixtral Large· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 93.8% |
| 5 | Qwen3.5 122B-A10B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 93.3% |
| 6 | Hunyuan Standard Vision· OpenVLM Leaderboard data | 93.2% |
| 7 | Mistral Small 3.2 24B Instruct2506· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 92.9% |
| 8 | Qwen3.5 27B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 92.9% |
| 9 | Qwen3.6 35B A3B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 92.7% |
| 10 | Qwen3.5 35B A3B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 92.6% |
| 11 | Llama 3.2 11B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 91.1% |
| 12 | InternVL3 78B· OpenVLM Leaderboard data | 89.8% |
| 13 | Qwen3 VL 235B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. | 89.7% |
| 14 | Gemini 2.5 Pro· OpenVLM Leaderboard data | 89.5% |
| 15 | GPT 5· OpenVLM Leaderboard data | 89.5% |
| 16 | Qwen3 VL 32B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. | 89.5% |
| 17 | Qwen3 VL 235Bthinking· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. | 89.2% |
| 18 | SenseNova-V6-Pro· OpenVLM Leaderboard data | 89.2% |
| 19 | Doubao Seed 1.8· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 89.1% |
| 20 | Qwen3 VL 32B Thinking· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. | 88.9% |
| 21 | internvl3-38b· OpenVLM Leaderboard data | 88.7% |
| 22 | Qwen2.5 VL 72B Instruct· OpenVLM Leaderboard dataValues from 2 sources
| 88.5% |
| 23 | Qwen2-VL-72B· OpenVLM Leaderboard data | 88.3% |
| 24 | GPT-4.5 Preview· OpenVLM Leaderboard data | 87.2% |
| 25 | qwen3-vl-30bthinking· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. | 86.9% |
| 26 | GPT-5 mini· OpenVLM Leaderboard data | 86.7% |
| 27 | ChatGPT-4o Latest· OpenVLM Leaderboard data | 86.3% |
| 28 | GPT 4.1· OpenVLM Leaderboard data | 85.9% |
| 29 | Qwen3 VL 8B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. | 85.7% |
| 30 | GLM 4 Vision Plus· OpenVLM Leaderboard data | 85.1% |
| 31 | internvl3-8b· OpenVLM Leaderboard data | 85.1% |
| 32 | qwen3-vl-30b· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. | 85% |
| 33 | Qwen: Qwen3 VL 8B Thinking· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. | 84.9% |
| 34 | Gemma 3 27B IT· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claimValues from 2 sources
| 84.5% |
| 35 | InternVL2 26B· OpenVLM Leaderboard data | 84.5% |
| 36 | gemini-1.5-flash-002· OpenVLM Leaderboard data | 84.4% |
| 37 | Qwen2.5-VL 7B Instruct· OpenVLM Leaderboard data | 84.3% |
| 38 | Gemma 3 12B IT· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claimValues from 2 sources
| 84.2% |
| 39 | DeepSeek VL2· OpenVLM Leaderboard dataValues from 2 sources
| 83.8% |
| 40 | Molmo 72B· OpenVLM Leaderboard data | 83.4% |
| 41 | gemini-1.5-pro-002· OpenVLM Leaderboard data | 83.3% |
| 42 | Qwen2.5-Omni 7B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 83.2% |
| 43 | Gemini 2.0 Flash· OpenVLM Leaderboard data | 83.1% |
| 44 | Phi 4 Multimodal Instruct· OpenVLM Leaderboard dataValues from 2 sources
| 83% |
| 45 | MiniCPM-V 2.6· OpenVLM Leaderboard data | 82.1% |
| 46 | Moonshot V1 8K· OpenVLM Leaderboard data | 81.8% |
| 47 | GPT-5 nano· OpenVLM Leaderboard data | 81.4% |
| 48 | qwen2.5-vl-3b-instruct· OpenVLM Leaderboard data | 81.4% |
| 49 | Grok 2 Vision 1212· OpenVLM Leaderboard data | 81.2% |
| 50 | Molmo 7B-D· OpenVLM Leaderboard data | 81% |
Showing 1–50 of 71 results.
Data credits: OpenVLM Leaderboard (OpenCompass / VLMEvalKit), Apache-2.0 (licence) · Data by LLM Stats