Benchmark Almanac

MMStar

Qwen 3.8 Max 85.9% leads since August 2026
GPT-5 75.7% leads since August 2025
Qwen 3.8 Max 85.9% leads since August 2026
In use

The official leaderboard hasn’t scored a model released after GPT-5 (August 2025). Labs report their own numbers for 30 newer models, such as Qwen 3.8 Max at 85.9%.

Models answer 1,500 human-curated image questions that cannot be solved without the image, across 6 core capabilities. Scored by accuracy.

Released
March 2024
Kind
General visual questions
Level
School
Scored
Checked automatically
Answers
Multiple choice
Tasks
All public
Measures
Capability
Language
English
In and out
Text, Images → Text
Interaction
Single turn
Refresh
One fixed set
Run by
Anyone
In views
Vision
Strongest model tested
Qwen 3.8 Max #30 on the capability index
Last in a launch post
October 2026
Last model added
October 2026
Scores refreshed
November 18th, 2026

In the authors’ words

Large vision-language models (LVLMs) have recently achieved rapid progress, sparking numerous studies to evaluate their multi-modal capabilities. However, we dig into current evaluation works and identify two primary issues: 1) Visual content is unnecessary for many samples. The answers can be directly inferred from the questions and options, or the world knowledge embedded in LLMs. This phenomenon is prevalent across current benchmarks. For instance, GeminiPro achieves 42.9% on the MMMU benchmark without any visual input, and outperforms the random choice baseline across six benchmarks over 24% on average. 2) Unintentional data leakage exists in LLM and LVLM training. LLM and LVLM could still answer some visual-necessary questions without visual content, indicating the memorizing of these samples within large-scale training data. For example, Sphinx-X-MoE gets 43.6% on MMMU without accessing images, surpassing its LLM backbone with 17.9%. Both problems lead to misjudgments of actual multi-modal gains and potentially misguide the study of LVLM. To this end, we present MMStar, an elite vision-indispensable multi-modal benchmark comprising 1,500 samples meticulously selected by humans. MMStar benchmarks 6 core capabilities and 18 detailed axes, aiming to evaluate LVLMs' multi-modal capacities with carefully balanced and purified samples. These samples are first roughly selected from current benchmarks with an automated pipeline, human review is then involved to ensure each curated sample exhibits visual dependency, minimal data leakage, and requires advanced multi-modal capabilities. Moreover, two metrics are developed to measure data leakage and actual performance gain in multi-modal training. We evaluate 16 leading LVLMs on MMStar to assess their multi-modal capabilities, and on 7 benchmarks with the proposed metrics to investigate their data leakage and actual multi-modal gain.

Abstract of Are We on the Right Way for Evaluating Large Vision-Language Models? (arXiv:2403.20330, CC0)
37
Models scored in lab cards
35
Launch posts list it, from 14 labs
22
System cards report it, from 13 labs

Qwen 3.8 Max leads with 85.9%

Best score per model as labs report them in their own cards, drawn striped. Labs pick their own settings, so these compare less cleanly than independent runs.

GPT-5 leads with 75.7% across all independent runs

Best score per model across the official leaderboard, Epoch AI’s own runs and any board Epoch AI copied. Dotted bars are reruns by Epoch AI or another board, and each row names its source.

Qwen 3.8 Max leads with 85.9%, claims included

Best score per model from every source, including labs’ own numbers and claims Epoch AI collected from reports and papers. Labs pick their own settings, so a claim can beat an independent run of the same model. Striped bars are claims, and dotted bars are reruns by Epoch AI or another board.

  1. 1 Qwen 3.8 Max 85.9%
  2. 2 Qwen3.5 397B-A17B 83.8%
  3. 3 Qwen 3.6 Plus 83.3%
  4. 4 Seed2.0 Pro 83%
  5. 5 Qwen3.5-122B-A10B 82.9%
  6. 6 Qwen3.5-35B-A3B 81.9%
  7. 7 Qwen3.6 27B 81.4%
  8. 8 Qwen3.5-27B 81%
  9. 9 MiniCPM-V-4.7-35B-A3B 80.7%
  10. 10 Seed1.8 79.9%
  11. 11 Qwen3.5-9B 79.7%
  12. 12 Qwen3.5-Omni-Plus 79.4%
  13. 13 Qwen3-VL-235B-A22B 78.7%
  14. 14 Qwen3.5-4B 78.3%
  15. 15 InternVL3.5-241B-A28B 77.9%
  16. 16 Seed1.5-VL 77.8%
  17. 17 STEP3-VL-10B 77.5%
  18. 18 GLM-4.6V 75.9%
  19. 19 Keye-VL-2.0-30B-A3B 75.9%
  20. 20 ERNIE 5.0 75.5%
  21. 21 GLM-4.5V 75.3%
  22. 22 Intern-S1 74.9%
  23. 23 EXAONE 4.5 33B 74.9%
  24. 24 GLM-4.1V-9B-Thinking 72.9%
  25. 25 Nemotron Nano 12B v2 VL 71.7%
  26. 26 Qwen3.5-2B 71.7%
  27. 27 ERNIE-4.5-VL-28B-A3B-Thinking 71.1%
  28. 28 LongCat-Flash-Omni 70.9%
  29. 29 Qwen2.5-VL-72B 70.8%
  30. 30 Kimi-VL-A3B-Thinking-2506 70.4%
  31. 31 Qwen3-Omni-30B-A3B 68.5%
  32. 32 Phi-4-Reasoning-Vision-15B 64.5%
  33. 33 Qwen2.5-Omni-7B 64%
  34. 34 LFM2.5-VL-3B 63.3%
  35. 35 Qwen3.5-0.8B 58.3%
  36. 36 North Micro Vision Instruct 51.8%
  37. 37 LFM2-VL-1.6B 49.9%

Source: Labs’ model cards

174 models. Scroll sideways for the rest.

  1. 75.7 GPT-5
  2. 73.6 Gemini 2.5 Pro
  3. 73.4 InternVL3-78B
  4. 73.2 GPT-5 mini
  5. 72.6 InternVL3-38B
  6. 72.1 InternVL2.5-78B-MPO
  7. 70.5 Qwen2.5-VL-72B
  8. 70.3 Qwen2.5-VL-32B
  9. 70.1 InternVL2.5-38B-MPO
  10. 70.0 Kimi-VL-A3B-Thinking-2506
  11. 69.8 GPT-4.1
  12. 69.5 InternVL2_5-78B
  13. 69.5 SAIL-VL-1.6-8B
  14. 69.4 Gemini 2.0 Flash
  15. 69.3 GPT-4.5
  16. 69.2 Ovis2-34B
  17. 68.9 InternVL3-14B
  18. 68.7 InternVL3-8B
  19. 68.6 Qwen2-VL-72B-Instruct
  20. 68.5 InternVL2.5-38B
  21. 68.5 Gemini 2.0 Pro
  22. 68.3 GPT-5 nano
  23. 67.7 InternVL2.5-26B-MPO
  24. 67.4 InternVL3-9B
  25. 67.2 Ovis2-16B
  26. 67.1 InternVL2-Llama3-76B
  27. 67.1 Gemini 1.5 Pro
  28. 66.5 InternVL2.5-26B
  29. 65.8 LLaVA-OneVision 72B
  30. 65.2 Grok 2 Vision (1212)
  31. 65.2 InternVL2.5-8B-MPO
  32. 65.1 Claude 3.5 Sonnet (October 2024)
  33. 65.1 GPT-4o
  34. 65.1 Claude 3.7 Sonnet
  35. 64.7 InternVL2-40B
  36. 64.7 GPT-4o
  37. 64.6 Ovis2-8B
  38. 64.4 Gemini 1.5 Flash
  39. 64.1 Qwen2.5-VL-7B
  40. 63.9 GPT-4o
  41. 63.7 NVLM-D-72B
  42. 63.5 Ovis1.6-Gemma2-27B
  43. 63.3 Molmo-72B
  44. 63.3 MiniCPM-o-2_6
  45. 63.2 InternVL2.5-8B
  46. 62.9 Ovis1.6-Gemma2-9B
  47. 62.5 InternVL2-8B-MPO
  48. 62.2 Claude 3.5 Sonnet
  49. 62.0 Kimi-VL-A3B-Instruct
  50. 61.9 LLaVA-OneVision 7B
  51. 61.9 DeepSeek-VL2
  52. 61.9 Ovis2-4B
  53. 61.5 InternVL2-8B
  54. 61.1 InternVL3-2B
  55. 61.0 InternVL2-26B
  56. 61.0 InternVL2.5-4B-MPO
  57. 60.9 GPT-4.1 mini
  58. 60.8 Kimi-VL-A3B-Thinking
  59. 60.7 Qwen2-VL-7B-Instruct
  60. 60.7 Aria
  61. 60.2 VITA-1.5
  62. 59.9 InternLM-XComposer2.5-7B
  63. 59.6 Gemma 3 27B
  64. 59.1 Gemini 1.5 Pro
  65. 58.9 Phi-4-multimodal
  66. 58.7 InternVL2.5-4B
  67. 58.1 Ovis1.5-Gemma2-9B
  68. 57.5 MiniCPM-V-2_6
  69. 57.5 DeepSeek-VL2-Small
  70. 57.3 Ovis1.5-Llama3-8B
  71. 57.1 InternVL-Chat-V1-5
  72. 56.7 LLaVA-OneVision 7B (SI)
  73. 56.7 Ovis2-2B
  74. 56.5 SAIL-VL-2B
  75. 56.3 Qwen2.5-VL-3B
  76. 56.1 Molmo-7B-D
  77. 56.1 Gemma 3 12B
  78. 56.0 GPT-4V
  79. 55.5 Eagle-X5-34B-Chat
  80. 55.3 InternLM-XComposer2-4KHD-7B
  81. 55.3 Llama 3.2 90B
  82. 55.2 VILA1.5-40B
  83. 55.0 Idefics3-8B-Llama3
  84. 54.9 Aquila-VL-2B
  85. 54.9 InternVL2.5-2B-MPO
  86. 54.8 GLM-4V-9B
  87. 54.8 GPT-4o mini
  88. 54.6 Granite Vision 3.2 2B
  89. 54.5 Pixtral 12B
  90. 54.3 InternVL2.5-2B
  91. 54.2 Cambrian-1-34B
  92. 53.9 InternVL2-4B
  93. 53.1 Mini-InternVL-Chat-4B-V1-5
  94. 52.3 InternVL3-1B
  95. 52.1 Ovis2-1B
  96. 52.0 Ovis1.6-Llama3.2-3B
  97. 51.9 Granite Vision 3.3 2B
  98. 51.8 MiniCPM-Llama3-V 2.5
  99. 51.6 LLaVA-v1.6-34B
  100. 51.3 InternVL2.5-1B
  101. 50.7 Cambrian-1-8B
  102. 50.5 CogVLM2-19B-Chat
  103. 50.2 Reka Flash
  104. 50.1 Molmo-7B-O
  105. 49.8 InternVL2-2B
  106. 49.8 Llama 3.2 11B
  107. 49.8 DeepSeek-VL2-Tiny
  108. 49.5 Qwen-VL-Max
  109. 49.5 Idefics2-8B
  110. 49.5 InternVL2.5-1B-MPO
  111. 48.3 PaliGemma-3B-mix-448
  112. 47.9 Gemma 3 4B
  113. 47.7 Phi-3-vision-128k-instruct
  114. 47.5 Phi-3.5-vision-instruct
  115. 47.5 Qwen2-VL-2B-Instruct
  116. 47.5 GPT-4.1 nano
  117. 47.1 Cambrian-1-13B
  118. 46.7 Mini-InternVL-Chat-2B-V1-5
  119. 46.6 VITA
  120. 46.6 Emu3-Chat
  121. 46.5 Janus-Pro-7B
  122. 46.4 InternLM-XComposer2-VL-1.8B
  123. 46.0 SmolVLM2-2.2B-Instruct
  124. 45.7 Claude 3 Opus
  125. 45.6 InternVL2-1B
  126. 45.4 Bunny-Llama-3-8B-V
  127. 45.4 Falcon2-11B-VLM
  128. 45.3 LLaVA-NeXT-Interleave-7B-DPO
  129. 44.5 LLaVA-NeXT-Interleave-7B
  130. 44.2 Claude 3 Sonnet
  131. 44.2 VILA1.5-13B
  132. 43.9 llama3-llava-next-8b
  133. 43.7 Eagle-X5-13B
  134. 43.1 MolmoE-1B
  135. 41.7 Eagle-X5-7B
  136. 41.7 SmolVLM-Instruct
  137. 40.7 Emu2-Chat
  138. 40.7 Monkey-Chat
  139. 40.6 VILA1.5-3B
  140. 40.5 Yi-VL-34B
  141. 40.5 DeepSeek-VL-7B-Chat
  142. 40.4 llava-v1.6-vicuna-13b
  143. 39.9 CogVLM-17B-Chat
  144. 39.9 DeepSeek-VL-1.3B-Chat
  145. 39.7 Qwen-VL-Plus
  146. 39.7 Llama-3-VILA1.5-8B
  147. 39.6 OmniLMM-12B
  148. 39.1 MiniCPM-V 2.0
  149. 38.6 Gemini 1.0 Pro
  150. 38.6 MiniCPM-V
  151. 38.4 llava-v1.6-mistral-7b
  152. 38.3 ShareGPT4V-13B
  153. 38.3 SmolVLM-500M-Instruct
  154. 38.2 SmolVLM2-500M-Video-Instruct
  155. 38.1 Claude 3 Haiku
  156. 37.7 Yi-VL-6B
  157. 37.7 LLaVA-OneVision 0.5B
  158. 37.6 llava-v1.6-vicuna-7b
  159. 37.6 Janus-1.3B
  160. 37.5 LLaVA-OneVision 0.5B (SI)
  161. 37.0 Monkey
  162. 35.7 ShareGPT4V-7B
  163. 34.8 mPLUG-Owl2
  164. 34.5 Qwen-VL-Chat
  165. 34.5 SmolVLM-256M-Instruct
  166. 34.3 LLaVA-v1.5-13B
  167. 33.1 llava-v1.5-7b
  168. 32.7 instructblip-vicuna-7b
  169. 31.8 Chameleon-30B
  170. 31.1 Chameleon-7B
  171. 30.9 SmolVLM2-256M-Video-Instruct
  172. 26.1 IDEFICS-80B-Instruct
  173. 24.9 Kosmos-2
  174. 21.6 IDEFICS-9B-Instruct
  1. 1 GPT-5 ✓ 75.7%
  2. 2 Gemini 2.5 Pro Mar 2025 ✓ 73.6%
  3. 3 InternVL3-78B ✓ 73.4%
  4. 4 GPT-5 mini ✓ 73.2%
  5. 5 InternVL3-38B ✓ 72.6%
  6. 6 InternVL2.5-78B-MPO ✓ 72.1%
  7. 7 Qwen2.5-VL-72B ✓ 70.5%
  8. 8 Qwen2.5-VL-32B ✓ 70.3%
  9. 9 InternVL2.5-38B-MPO ✓ 70.1%
  10. 10 Kimi-VL-A3B-Thinking-2506 ✓ 70%
  11. 11 GPT-4.1 ✓ 69.8%
  12. 12 InternVL2_5-78B ✓ 69.5%
  13. 13 SAIL-VL-1.6-8B ✓ 69.5%
  14. 14 Gemini 2.0 Flash Dec 2024 ✓ 69.4%
  15. 15 GPT-4.5 ✓ 69.3%
  16. 16 Ovis2-34B ✓ 69.2%
  17. 17 InternVL3-14B ✓ 68.9%
  18. 18 InternVL3-8B ✓ 68.7%
  19. 19 Qwen2-VL-72B-Instruct ✓ 68.6%
  20. 20 InternVL2.5-38B ✓ 68.5%
  21. 21 Gemini 2.0 Pro ✓ 68.5%
  22. 22 GPT-5 nano ✓ 68.3%
  23. 23 InternVL2.5-26B-MPO ✓ 67.7%
  24. 24 InternVL3-9B ✓ 67.4%
  25. 25 Ovis2-16B ✓ 67.2%
  26. 26 InternVL2-Llama3-76B ✓ 67.1%
  27. 27 Gemini 1.5 Pro Sep 2024 ✓ 67.1%
  28. 28 InternVL2.5-26B ✓ 66.5%
  29. 29 LLaVA-OneVision 72B ✓ 65.8%
  30. 30 Grok 2 Vision (1212) ✓ 65.2%
  31. 31 InternVL2.5-8B-MPO ✓ 65.2%
  32. 32 Claude 3.5 Sonnet (October 2024) ✓ 65.1%
  33. 33 GPT-4o Nov 2024, detail-high ✓ 65.1%
  34. 34 Claude 3.7 Sonnet ✓ 65.1%
  35. 35 InternVL2-40B ✓ 64.7%
  36. 36 GPT-4o Aug 2024, detail-high ✓ 64.7%
  37. 37 Ovis2-8B ✓ 64.6%
  38. 38 Gemini 1.5 Flash Sep 2024 ✓ 64.4%
  39. 39 Qwen2.5-VL-7B ✓ 64.1%
  40. 40 GPT-4o May 2024, detail-high ✓ 63.9%
  41. 41 NVLM-D-72B ✓ 63.7%
  42. 42 Ovis1.6-Gemma2-27B ✓ 63.5%
  43. 43 Molmo-72B ✓ 63.3%
  44. 44 MiniCPM-o-2_6 ✓ 63.3%
  45. 45 InternVL2.5-8B ✓ 63.2%
  46. 46 Ovis1.6-Gemma2-9B ✓ 62.9%
  47. 47 InternVL2-8B-MPO ✓ 62.5%
  48. 48 Claude 3.5 Sonnet ✓ 62.2%
  49. 49 Kimi-VL-A3B-Instruct ✓ 62%
  50. 50 LLaVA-OneVision 7B ✓ 61.9%
  51. 51 DeepSeek-VL2 ✓ 61.9%
  52. 52 Ovis2-4B ✓ 61.9%
  53. 53 InternVL2-8B ✓ 61.5%
  54. 54 InternVL3-2B ✓ 61.1%
  55. 55 InternVL2-26B ✓ 61%
  56. 56 InternVL2.5-4B-MPO ✓ 61%
  57. 57 GPT-4.1 mini ✓ 60.9%
  58. 58 Kimi-VL-A3B-Thinking ✓ 60.8%
  59. 59 Qwen2-VL-7B-Instruct ✓ 60.7%
  60. 60 Aria ✓ 60.7%
  61. 61 VITA-1.5 ✓ 60.2%
  62. 62 InternLM-XComposer2.5-7B ✓ 59.9%
  63. 63 Gemma 3 27B ✓ 59.6%
  64. 64 Gemini 1.5 Pro May 2024 ✓ 59.1%
  65. 65 Phi-4-multimodal ✓ 58.9%
  66. 66 InternVL2.5-4B ✓ 58.7%
  67. 67 Ovis1.5-Gemma2-9B ✓ 58.1%
  68. 68 MiniCPM-V-2_6 ✓ 57.5%
  69. 69 DeepSeek-VL2-Small ✓ 57.5%
  70. 70 Ovis1.5-Llama3-8B ✓ 57.3%
  71. 71 InternVL-Chat-V1-5 ✓ 57.1%
  72. 72 LLaVA-OneVision 7B (SI) ✓ 56.7%
  73. 73 Ovis2-2B ✓ 56.7%
  74. 74 SAIL-VL-2B ✓ 56.5%
  75. 75 Qwen2.5-VL-3B ✓ 56.3%
  76. 76 Molmo-7B-D ✓ 56.1%
  77. 77 Gemma 3 12B ✓ 56.1%
  78. 78 GPT-4V detail-high ✓ 56%
  79. 79 Eagle-X5-34B-Chat ✓ 55.5%
  80. 80 InternLM-XComposer2-4KHD-7B ✓ 55.3%
  81. 81 Llama 3.2 90B ✓ 55.3%
  82. 82 VILA1.5-40B ✓ 55.2%
  83. 83 Idefics3-8B-Llama3 ✓ 55%
  84. 84 Aquila-VL-2B ✓ 54.9%
  85. 85 InternVL2.5-2B-MPO ✓ 54.9%
  86. 86 GLM-4V-9B ✓ 54.8%
  87. 87 GPT-4o mini detail-high ✓ 54.8%
  88. 88 Granite Vision 3.2 2B ✓ 54.6%
  89. 89 Pixtral 12B ✓ 54.5%
  90. 90 InternVL2.5-2B ✓ 54.3%
  91. 91 Cambrian-1-34B ✓ 54.2%
  92. 92 InternVL2-4B ✓ 53.9%
  93. 93 Mini-InternVL-Chat-4B-V1-5 ✓ 53.1%
  94. 94 InternVL3-1B ✓ 52.3%
  95. 95 Ovis2-1B ✓ 52.1%
  96. 96 Ovis1.6-Llama3.2-3B ✓ 52%
  97. 97 Granite Vision 3.3 2B ✓ 51.9%
  98. 98 MiniCPM-Llama3-V 2.5 ✓ 51.8%
  99. 99 LLaVA-v1.6-34B ✓ 51.6%
  100. 100 InternVL2.5-1B ✓ 51.3%
  101. 101 Cambrian-1-8B ✓ 50.7%
  102. 102 CogVLM2-19B-Chat ✓ 50.5%
  103. 103 Reka Flash ✓ 50.2%
  104. 104 Molmo-7B-O ✓ 50.1%
  105. 105 InternVL2-2B ✓ 49.8%
  106. 106 Llama 3.2 11B ✓ 49.8%
  107. 107 DeepSeek-VL2-Tiny ✓ 49.8%
  108. 108 Qwen-VL-Max ✓ 49.5%
  109. 109 Idefics2-8B ✓ 49.5%
  110. 110 InternVL2.5-1B-MPO ✓ 49.5%
  111. 111 PaliGemma-3B-mix-448 ✓ 48.3%
  112. 112 Gemma 3 4B ✓ 47.9%
  113. 113 Phi-3-vision-128k-instruct ✓ 47.7%
  114. 114 Phi-3.5-vision-instruct ✓ 47.5%
  115. 115 Qwen2-VL-2B-Instruct ✓ 47.5%
  116. 116 GPT-4.1 nano ✓ 47.5%
  117. 117 Cambrian-1-13B ✓ 47.1%
  118. 118 Mini-InternVL-Chat-2B-V1-5 ✓ 46.7%
  119. 119 VITA ✓ 46.6%
  120. 120 Emu3-Chat ✓ 46.6%
  121. 121 Janus-Pro-7B ✓ 46.5%
  122. 122 InternLM-XComposer2-VL-1.8B ✓ 46.4%
  123. 123 SmolVLM2-2.2B-Instruct ✓ 46%
  124. 124 Claude 3 Opus ✓ 45.7%
  125. 125 InternVL2-1B ✓ 45.6%
  126. 126 Bunny-Llama-3-8B-V ✓ 45.4%
  127. 127 Falcon2-11B-VLM ✓ 45.4%
  128. 128 LLaVA-NeXT-Interleave-7B-DPO ✓ 45.3%
  129. 129 LLaVA-NeXT-Interleave-7B ✓ 44.5%
  130. 130 Claude 3 Sonnet ✓ 44.2%
  131. 131 VILA1.5-13B ✓ 44.2%
  132. 132 llama3-llava-next-8b ✓ 43.9%
  133. 133 Eagle-X5-13B ✓ 43.7%
  134. 134 MolmoE-1B ✓ 43.1%
  135. 135 Eagle-X5-7B ✓ 41.7%
  136. 136 SmolVLM-Instruct ✓ 41.7%
  137. 137 Emu2-Chat ✓ 40.7%
  138. 138 Monkey-Chat ✓ 40.7%
  139. 139 VILA1.5-3B ✓ 40.6%
  140. 140 Yi-VL-34B ✓ 40.5%
  141. 141 DeepSeek-VL-7B-Chat ✓ 40.5%
  142. 142 llava-v1.6-vicuna-13b ✓ 40.4%
  143. 143 CogVLM-17B-Chat ✓ 39.9%
  144. 144 DeepSeek-VL-1.3B-Chat ✓ 39.9%
  145. 145 Qwen-VL-Plus ✓ 39.7%
  146. 146 Llama-3-VILA1.5-8B ✓ 39.7%
  147. 147 OmniLMM-12B ✓ 39.6%
  148. 148 MiniCPM-V 2.0 ✓ 39.1%
  149. 149 Gemini 1.0 Pro ✓ 38.6%
  150. 150 MiniCPM-V ✓ 38.6%
  151. 151 llava-v1.6-mistral-7b ✓ 38.4%
  152. 152 ShareGPT4V-13B ✓ 38.3%
  153. 153 SmolVLM-500M-Instruct ✓ 38.3%
  154. 154 SmolVLM2-500M-Video-Instruct ✓ 38.2%
  155. 155 Claude 3 Haiku ✓ 38.1%
  156. 156 Yi-VL-6B ✓ 37.7%
  157. 157 LLaVA-OneVision 0.5B ✓ 37.7%
  158. 158 llava-v1.6-vicuna-7b ✓ 37.6%
  159. 159 Janus-1.3B ✓ 37.6%
  160. 160 LLaVA-OneVision 0.5B (SI) ✓ 37.5%
  161. 161 Monkey ✓ 37%
  162. 162 ShareGPT4V-7B ✓ 35.7%
  163. 163 mPLUG-Owl2 ✓ 34.8%
  164. 164 Qwen-VL-Chat ✓ 34.5%
  165. 165 SmolVLM-256M-Instruct ✓ 34.5%
  166. 166 LLaVA-v1.5-13B ✓ 34.3%
  167. 167 llava-v1.5-7b ✓ 33.1%
  168. 168 instructblip-vicuna-7b ✓ 32.7%
  169. 169 Chameleon-30B ✓ 31.8%
  170. 170 Chameleon-7B ✓ 31.1%
  171. 171 SmolVLM2-256M-Video-Instruct ✓ 30.9%
  172. 172 IDEFICS-80B-Instruct ✓ 26.1%
  173. 173 Kosmos-2 ✓ 24.9%
  174. 174 IDEFICS-9B-Instruct ✓ 21.6%

Source: Official MMStar leaderboard

209 models. Scroll sideways for the rest.

  1. 85.9 Qwen 3.8 Max
  2. 83.8 Qwen3.5 397B-A17B
  3. 83.3 Qwen 3.6 Plus
  4. 83.0 Seed2.0 Pro
  5. 82.9 Qwen3.5-122B-A10B
  6. 81.9 Qwen3.5-35B-A3B
  7. 81.4 Qwen3.6 27B
  8. 81.0 Qwen3.5-27B
  9. 80.7 MiniCPM-V-4.7-35B-A3B
  10. 79.9 Seed1.8
  11. 79.7 Qwen3.5-9B
  12. 79.4 Qwen3.5-Omni-Plus
  13. 78.7 Qwen3-VL-235B-A22B
  14. 78.3 Qwen3.5-4B
  15. 77.9 InternVL3.5-241B-A28B
  16. 77.8 Seed1.5-VL
  17. 77.5 STEP3-VL-10B
  18. 75.9 GLM-4.6V
  19. 75.9 Keye-VL-2.0-30B-A3B
  20. 75.7 GPT-5
  21. 75.5 ERNIE 5.0
  22. 75.3 GLM-4.5V
  23. 74.9 Intern-S1
  24. 74.9 EXAONE 4.5 33B
  25. 73.6 Gemini 2.5 Pro
  26. 73.4 InternVL3-78B
  27. 73.2 GPT-5 mini
  28. 72.9 GLM-4.1V-9B-Thinking
  29. 72.6 InternVL3-38B
  30. 72.1 InternVL2.5-78B-MPO
  31. 71.7 Nemotron Nano 12B v2 VL
  32. 71.7 Qwen3.5-2B
  33. 71.1 ERNIE-4.5-VL-28B-A3B-Thinking
  34. 70.9 LongCat-Flash-Omni
  35. 70.8 Qwen2.5-VL-72B
  36. 70.4 Kimi-VL-A3B-Thinking-2506
  37. 70.3 Qwen2.5-VL-32B
  38. 70.1 InternVL2.5-38B-MPO
  39. 69.8 GPT-4.1
  40. 69.5 InternVL2_5-78B
  41. 69.5 SAIL-VL-1.6-8B
  42. 69.4 Gemini 2.0 Flash
  43. 69.3 GPT-4.5
  44. 69.2 Ovis2-34B
  45. 68.9 InternVL3-14B
  46. 68.7 InternVL3-8B
  47. 68.6 Qwen2-VL-72B-Instruct
  48. 68.5 InternVL2.5-38B
  49. 68.5 Gemini 2.0 Pro
  50. 68.5 Qwen3-Omni-30B-A3B
  51. 68.3 GPT-5 nano
  52. 67.7 InternVL2.5-26B-MPO
  53. 67.4 InternVL3-9B
  54. 67.2 Ovis2-16B
  55. 67.1 InternVL2-Llama3-76B
  56. 67.1 Gemini 1.5 Pro
  57. 66.5 InternVL2.5-26B
  58. 65.8 LLaVA-OneVision 72B
  59. 65.2 Grok 2 Vision (1212)
  60. 65.2 InternVL2.5-8B-MPO
  61. 65.1 Claude 3.5 Sonnet (October 2024)
  62. 65.1 GPT-4o
  63. 65.1 Claude 3.7 Sonnet
  64. 64.7 InternVL2-40B
  65. 64.7 GPT-4o
  66. 64.6 Ovis2-8B
  67. 64.5 Phi-4-Reasoning-Vision-15B
  68. 64.4 Gemini 1.5 Flash
  69. 64.1 Qwen2.5-VL-7B
  70. 64.0 Qwen2.5-Omni-7B
  71. 63.9 GPT-4o
  72. 63.7 NVLM-D-72B
  73. 63.5 Ovis1.6-Gemma2-27B
  74. 63.3 Molmo-72B
  75. 63.3 MiniCPM-o-2_6
  76. 63.3 LFM2.5-VL-3B
  77. 63.2 InternVL2.5-8B
  78. 62.9 Ovis1.6-Gemma2-9B
  79. 62.5 InternVL2-8B-MPO
  80. 62.2 Claude 3.5 Sonnet
  81. 62.0 Kimi-VL-A3B-Instruct
  82. 61.9 LLaVA-OneVision 7B
  83. 61.9 DeepSeek-VL2
  84. 61.9 Ovis2-4B
  85. 61.5 InternVL2-8B
  86. 61.1 InternVL3-2B
  87. 61.0 InternVL2-26B
  88. 61.0 InternVL2.5-4B-MPO
  89. 60.9 GPT-4.1 mini
  90. 60.8 Kimi-VL-A3B-Thinking
  91. 60.7 Qwen2-VL-7B-Instruct
  92. 60.7 Aria
  93. 60.2 VITA-1.5
  94. 59.9 InternLM-XComposer2.5-7B
  95. 59.6 Gemma 3 27B
  96. 59.1 Gemini 1.5 Pro
  97. 58.9 Phi-4-multimodal
  98. 58.7 InternVL2.5-4B
  99. 58.3 Qwen3.5-0.8B
  100. 58.1 Ovis1.5-Gemma2-9B
  101. 57.5 MiniCPM-V-2_6
  102. 57.5 DeepSeek-VL2-Small
  103. 57.3 Ovis1.5-Llama3-8B
  104. 57.1 InternVL-Chat-V1-5
  105. 56.7 LLaVA-OneVision 7B (SI)
  106. 56.7 Ovis2-2B
  107. 56.5 SAIL-VL-2B
  108. 56.3 Qwen2.5-VL-3B
  109. 56.1 Molmo-7B-D
  110. 56.1 Gemma 3 12B
  111. 56.0 GPT-4V
  112. 55.5 Eagle-X5-34B-Chat
  113. 55.3 InternLM-XComposer2-4KHD-7B
  114. 55.3 Llama 3.2 90B
  115. 55.2 VILA1.5-40B
  116. 55.0 Idefics3-8B-Llama3
  117. 54.9 Aquila-VL-2B
  118. 54.9 InternVL2.5-2B-MPO
  119. 54.8 GLM-4V-9B
  120. 54.8 GPT-4o mini
  121. 54.6 Granite Vision 3.2 2B
  122. 54.5 Pixtral 12B
  123. 54.3 InternVL2.5-2B
  124. 54.2 Cambrian-1-34B
  125. 53.9 InternVL2-4B
  126. 53.1 Mini-InternVL-Chat-4B-V1-5
  127. 52.3 InternVL3-1B
  128. 52.1 Ovis2-1B
  129. 52.0 Ovis1.6-Llama3.2-3B
  130. 51.9 Granite Vision 3.3 2B
  131. 51.8 MiniCPM-Llama3-V 2.5
  132. 51.8 North Micro Vision Instruct
  133. 51.6 LLaVA-v1.6-34B
  134. 51.3 InternVL2.5-1B
  135. 50.7 Cambrian-1-8B
  136. 50.5 CogVLM2-19B-Chat
  137. 50.2 Reka Flash
  138. 50.1 Molmo-7B-O
  139. 49.9 LFM2-VL-1.6B
  140. 49.8 InternVL2-2B
  141. 49.8 Llama 3.2 11B
  142. 49.8 DeepSeek-VL2-Tiny
  143. 49.5 Qwen-VL-Max
  144. 49.5 Idefics2-8B
  145. 49.5 InternVL2.5-1B-MPO
  146. 48.3 PaliGemma-3B-mix-448
  147. 47.9 Gemma 3 4B
  148. 47.7 Phi-3-vision-128k-instruct
  149. 47.5 Phi-3.5-vision-instruct
  150. 47.5 Qwen2-VL-2B-Instruct
  151. 47.5 GPT-4.1 nano
  152. 47.1 Cambrian-1-13B
  153. 46.7 Mini-InternVL-Chat-2B-V1-5
  154. 46.6 VITA
  155. 46.6 Emu3-Chat
  156. 46.5 Janus-Pro-7B
  157. 46.4 InternLM-XComposer2-VL-1.8B
  158. 46.0 SmolVLM2-2.2B-Instruct
  159. 45.7 Claude 3 Opus
  160. 45.6 InternVL2-1B
  161. 45.4 Bunny-Llama-3-8B-V
  162. 45.4 Falcon2-11B-VLM
  163. 45.3 LLaVA-NeXT-Interleave-7B-DPO
  164. 44.5 LLaVA-NeXT-Interleave-7B
  165. 44.2 Claude 3 Sonnet
  166. 44.2 VILA1.5-13B
  167. 43.9 llama3-llava-next-8b
  168. 43.7 Eagle-X5-13B
  169. 43.1 MolmoE-1B
  170. 41.7 Eagle-X5-7B
  171. 41.7 SmolVLM-Instruct
  172. 40.7 Emu2-Chat
  173. 40.7 Monkey-Chat
  174. 40.6 VILA1.5-3B
  175. 40.5 Yi-VL-34B
  176. 40.5 DeepSeek-VL-7B-Chat
  177. 40.4 llava-v1.6-vicuna-13b
  178. 39.9 CogVLM-17B-Chat
  179. 39.9 DeepSeek-VL-1.3B-Chat
  180. 39.7 Qwen-VL-Plus
  181. 39.7 Llama-3-VILA1.5-8B
  182. 39.6 OmniLMM-12B
  183. 39.1 MiniCPM-V 2.0
  184. 38.6 Gemini 1.0 Pro
  185. 38.6 MiniCPM-V
  186. 38.4 llava-v1.6-mistral-7b
  187. 38.3 ShareGPT4V-13B
  188. 38.3 SmolVLM-500M-Instruct
  189. 38.2 SmolVLM2-500M-Video-Instruct
  190. 38.1 Claude 3 Haiku
  191. 37.7 Yi-VL-6B
  192. 37.7 LLaVA-OneVision 0.5B
  193. 37.6 llava-v1.6-vicuna-7b
  194. 37.6 Janus-1.3B
  195. 37.5 LLaVA-OneVision 0.5B (SI)
  196. 37.0 Monkey
  197. 35.7 ShareGPT4V-7B
  198. 34.8 mPLUG-Owl2
  199. 34.5 Qwen-VL-Chat
  200. 34.5 SmolVLM-256M-Instruct
  201. 34.3 LLaVA-v1.5-13B
  202. 33.1 llava-v1.5-7b
  203. 32.7 instructblip-vicuna-7b
  204. 31.8 Chameleon-30B
  205. 31.1 Chameleon-7B
  206. 30.9 SmolVLM2-256M-Video-Instruct
  207. 26.1 IDEFICS-80B-Instruct
  208. 24.9 Kosmos-2
  209. 21.6 IDEFICS-9B-Instruct
  1. 1 Qwen 3.8 Max 85.9%
  2. 2 Qwen3.5 397B-A17B 83.8%
  3. 3 Qwen 3.6 Plus 83.3%
  4. 4 Seed2.0 Pro 83%
  5. 5 Qwen3.5-122B-A10B 82.9%
  6. 6 Qwen3.5-35B-A3B 81.9%
  7. 7 Qwen3.6 27B 81.4%
  8. 8 Qwen3.5-27B 81%
  9. 9 MiniCPM-V-4.7-35B-A3B 80.7%
  10. 10 Seed1.8 79.9%
  11. 11 Qwen3.5-9B 79.7%
  12. 12 Qwen3.5-Omni-Plus 79.4%
  13. 13 Qwen3-VL-235B-A22B 78.7%
  14. 14 Qwen3.5-4B 78.3%
  15. 15 InternVL3.5-241B-A28B 77.9%
  16. 16 Seed1.5-VL 77.8%
  17. 17 STEP3-VL-10B 77.5%
  18. 18 GLM-4.6V 75.9%
  19. 19 Keye-VL-2.0-30B-A3B 75.9%
  20. 20 GPT-5 ✓ 75.7%
  21. 21 ERNIE 5.0 75.5%
  22. 22 GLM-4.5V 75.3%
  23. 23 Intern-S1 74.9%
  24. 24 EXAONE 4.5 33B 74.9%
  25. 25 Gemini 2.5 Pro Mar 2025 ✓ 73.6%
  26. 26 InternVL3-78B ✓ 73.4%
  27. 27 GPT-5 mini ✓ 73.2%
  28. 28 GLM-4.1V-9B-Thinking 72.9%
  29. 29 InternVL3-38B ✓ 72.6%
  30. 30 InternVL2.5-78B-MPO ✓ 72.1%
  31. 31 Nemotron Nano 12B v2 VL 71.7%
  32. 32 Qwen3.5-2B 71.7%
  33. 33 ERNIE-4.5-VL-28B-A3B-Thinking 71.1%
  34. 34 LongCat-Flash-Omni 70.9%
  35. 35 Qwen2.5-VL-72B 70.8%
  36. 36 Kimi-VL-A3B-Thinking-2506 70.4%
  37. 37 Qwen2.5-VL-32B ✓ 70.3%
  38. 38 InternVL2.5-38B-MPO ✓ 70.1%
  39. 39 GPT-4.1 ✓ 69.8%
  40. 40 InternVL2_5-78B ✓ 69.5%
  41. 41 SAIL-VL-1.6-8B ✓ 69.5%
  42. 42 Gemini 2.0 Flash Dec 2024 ✓ 69.4%
  43. 43 GPT-4.5 ✓ 69.3%
  44. 44 Ovis2-34B ✓ 69.2%
  45. 45 InternVL3-14B ✓ 68.9%
  46. 46 InternVL3-8B ✓ 68.7%
  47. 47 Qwen2-VL-72B-Instruct ✓ 68.6%
  48. 48 InternVL2.5-38B ✓ 68.5%
  49. 49 Gemini 2.0 Pro ✓ 68.5%
  50. 50 Qwen3-Omni-30B-A3B 68.5%
  51. 51 GPT-5 nano ✓ 68.3%
  52. 52 InternVL2.5-26B-MPO ✓ 67.7%
  53. 53 InternVL3-9B ✓ 67.4%
  54. 54 Ovis2-16B ✓ 67.2%
  55. 55 InternVL2-Llama3-76B ✓ 67.1%
  56. 56 Gemini 1.5 Pro Sep 2024 ✓ 67.1%
  57. 57 InternVL2.5-26B ✓ 66.5%
  58. 58 LLaVA-OneVision 72B ✓ 65.8%
  59. 59 Grok 2 Vision (1212) ✓ 65.2%
  60. 60 InternVL2.5-8B-MPO ✓ 65.2%
  61. 61 Claude 3.5 Sonnet (October 2024) ✓ 65.1%
  62. 62 GPT-4o Nov 2024, detail-high ✓ 65.1%
  63. 63 Claude 3.7 Sonnet ✓ 65.1%
  64. 64 InternVL2-40B ✓ 64.7%
  65. 65 GPT-4o Aug 2024, detail-high ✓ 64.7%
  66. 66 Ovis2-8B ✓ 64.6%
  67. 67 Phi-4-Reasoning-Vision-15B 64.5%
  68. 68 Gemini 1.5 Flash Sep 2024 ✓ 64.4%
  69. 69 Qwen2.5-VL-7B ✓ 64.1%
  70. 70 Qwen2.5-Omni-7B 64%
  71. 71 GPT-4o May 2024, detail-high ✓ 63.9%
  72. 72 NVLM-D-72B ✓ 63.7%
  73. 73 Ovis1.6-Gemma2-27B ✓ 63.5%
  74. 74 Molmo-72B ✓ 63.3%
  75. 75 MiniCPM-o-2_6 ✓ 63.3%
  76. 76 LFM2.5-VL-3B 63.3%
  77. 77 InternVL2.5-8B ✓ 63.2%
  78. 78 Ovis1.6-Gemma2-9B ✓ 62.9%
  79. 79 InternVL2-8B-MPO ✓ 62.5%
  80. 80 Claude 3.5 Sonnet ✓ 62.2%
  81. 81 Kimi-VL-A3B-Instruct ✓ 62%
  82. 82 LLaVA-OneVision 7B ✓ 61.9%
  83. 83 DeepSeek-VL2 ✓ 61.9%
  84. 84 Ovis2-4B ✓ 61.9%
  85. 85 InternVL2-8B ✓ 61.5%
  86. 86 InternVL3-2B ✓ 61.1%
  87. 87 InternVL2-26B ✓ 61%
  88. 88 InternVL2.5-4B-MPO ✓ 61%
  89. 89 GPT-4.1 mini ✓ 60.9%
  90. 90 Kimi-VL-A3B-Thinking ✓ 60.8%
  91. 91 Qwen2-VL-7B-Instruct ✓ 60.7%
  92. 92 Aria ✓ 60.7%
  93. 93 VITA-1.5 ✓ 60.2%
  94. 94 InternLM-XComposer2.5-7B ✓ 59.9%
  95. 95 Gemma 3 27B ✓ 59.6%
  96. 96 Gemini 1.5 Pro May 2024 ✓ 59.1%
  97. 97 Phi-4-multimodal ✓ 58.9%
  98. 98 InternVL2.5-4B ✓ 58.7%
  99. 99 Qwen3.5-0.8B 58.3%
  100. 100 Ovis1.5-Gemma2-9B ✓ 58.1%
  101. 101 MiniCPM-V-2_6 ✓ 57.5%
  102. 102 DeepSeek-VL2-Small ✓ 57.5%
  103. 103 Ovis1.5-Llama3-8B ✓ 57.3%
  104. 104 InternVL-Chat-V1-5 ✓ 57.1%
  105. 105 LLaVA-OneVision 7B (SI) ✓ 56.7%
  106. 106 Ovis2-2B ✓ 56.7%
  107. 107 SAIL-VL-2B ✓ 56.5%
  108. 108 Qwen2.5-VL-3B ✓ 56.3%
  109. 109 Molmo-7B-D ✓ 56.1%
  110. 110 Gemma 3 12B ✓ 56.1%
  111. 111 GPT-4V detail-high ✓ 56%
  112. 112 Eagle-X5-34B-Chat ✓ 55.5%
  113. 113 InternLM-XComposer2-4KHD-7B ✓ 55.3%
  114. 114 Llama 3.2 90B ✓ 55.3%
  115. 115 VILA1.5-40B ✓ 55.2%
  116. 116 Idefics3-8B-Llama3 ✓ 55%
  117. 117 Aquila-VL-2B ✓ 54.9%
  118. 118 InternVL2.5-2B-MPO ✓ 54.9%
  119. 119 GLM-4V-9B ✓ 54.8%
  120. 120 GPT-4o mini detail-high ✓ 54.8%
  121. 121 Granite Vision 3.2 2B ✓ 54.6%
  122. 122 Pixtral 12B ✓ 54.5%
  123. 123 InternVL2.5-2B ✓ 54.3%
  124. 124 Cambrian-1-34B ✓ 54.2%
  125. 125 InternVL2-4B ✓ 53.9%
  126. 126 Mini-InternVL-Chat-4B-V1-5 ✓ 53.1%
  127. 127 InternVL3-1B ✓ 52.3%
  128. 128 Ovis2-1B ✓ 52.1%
  129. 129 Ovis1.6-Llama3.2-3B ✓ 52%
  130. 130 Granite Vision 3.3 2B ✓ 51.9%
  131. 131 MiniCPM-Llama3-V 2.5 ✓ 51.8%
  132. 132 North Micro Vision Instruct 51.8%
  133. 133 LLaVA-v1.6-34B ✓ 51.6%
  134. 134 InternVL2.5-1B ✓ 51.3%
  135. 135 Cambrian-1-8B ✓ 50.7%
  136. 136 CogVLM2-19B-Chat ✓ 50.5%
  137. 137 Reka Flash ✓ 50.2%
  138. 138 Molmo-7B-O ✓ 50.1%
  139. 139 LFM2-VL-1.6B 49.9%
  140. 140 InternVL2-2B ✓ 49.8%
  141. 141 Llama 3.2 11B ✓ 49.8%
  142. 142 DeepSeek-VL2-Tiny ✓ 49.8%
  143. 143 Qwen-VL-Max ✓ 49.5%
  144. 144 Idefics2-8B ✓ 49.5%
  145. 145 InternVL2.5-1B-MPO ✓ 49.5%
  146. 146 PaliGemma-3B-mix-448 ✓ 48.3%
  147. 147 Gemma 3 4B ✓ 47.9%
  148. 148 Phi-3-vision-128k-instruct ✓ 47.7%
  149. 149 Phi-3.5-vision-instruct ✓ 47.5%
  150. 150 Qwen2-VL-2B-Instruct ✓ 47.5%
  151. 151 GPT-4.1 nano ✓ 47.5%
  152. 152 Cambrian-1-13B ✓ 47.1%
  153. 153 Mini-InternVL-Chat-2B-V1-5 ✓ 46.7%
  154. 154 VITA ✓ 46.6%
  155. 155 Emu3-Chat ✓ 46.6%
  156. 156 Janus-Pro-7B ✓ 46.5%
  157. 157 InternLM-XComposer2-VL-1.8B ✓ 46.4%
  158. 158 SmolVLM2-2.2B-Instruct ✓ 46%
  159. 159 Claude 3 Opus ✓ 45.7%
  160. 160 InternVL2-1B ✓ 45.6%
  161. 161 Bunny-Llama-3-8B-V ✓ 45.4%
  162. 162 Falcon2-11B-VLM ✓ 45.4%
  163. 163 LLaVA-NeXT-Interleave-7B-DPO ✓ 45.3%
  164. 164 LLaVA-NeXT-Interleave-7B ✓ 44.5%
  165. 165 Claude 3 Sonnet ✓ 44.2%
  166. 166 VILA1.5-13B ✓ 44.2%
  167. 167 llama3-llava-next-8b ✓ 43.9%
  168. 168 Eagle-X5-13B ✓ 43.7%
  169. 169 MolmoE-1B ✓ 43.1%
  170. 170 Eagle-X5-7B ✓ 41.7%
  171. 171 SmolVLM-Instruct ✓ 41.7%
  172. 172 Emu2-Chat ✓ 40.7%
  173. 173 Monkey-Chat ✓ 40.7%
  174. 174 VILA1.5-3B ✓ 40.6%
  175. 175 Yi-VL-34B ✓ 40.5%
  176. 176 DeepSeek-VL-7B-Chat ✓ 40.5%
  177. 177 llava-v1.6-vicuna-13b ✓ 40.4%
  178. 178 CogVLM-17B-Chat ✓ 39.9%
  179. 179 DeepSeek-VL-1.3B-Chat ✓ 39.9%
  180. 180 Qwen-VL-Plus ✓ 39.7%
  181. 181 Llama-3-VILA1.5-8B ✓ 39.7%
  182. 182 OmniLMM-12B ✓ 39.6%
  183. 183 MiniCPM-V 2.0 ✓ 39.1%
  184. 184 Gemini 1.0 Pro ✓ 38.6%
  185. 185 MiniCPM-V ✓ 38.6%
  186. 186 llava-v1.6-mistral-7b ✓ 38.4%
  187. 187 ShareGPT4V-13B ✓ 38.3%
  188. 188 SmolVLM-500M-Instruct ✓ 38.3%
  189. 189 SmolVLM2-500M-Video-Instruct ✓ 38.2%
  190. 190 Claude 3 Haiku ✓ 38.1%
  191. 191 Yi-VL-6B ✓ 37.7%
  192. 192 LLaVA-OneVision 0.5B ✓ 37.7%
  193. 193 llava-v1.6-vicuna-7b ✓ 37.6%
  194. 194 Janus-1.3B ✓ 37.6%
  195. 195 LLaVA-OneVision 0.5B (SI) ✓ 37.5%
  196. 196 Monkey ✓ 37%
  197. 197 ShareGPT4V-7B ✓ 35.7%
  198. 198 mPLUG-Owl2 ✓ 34.8%
  199. 199 Qwen-VL-Chat ✓ 34.5%
  200. 200 SmolVLM-256M-Instruct ✓ 34.5%
  201. 201 LLaVA-v1.5-13B ✓ 34.3%
  202. 202 llava-v1.5-7b ✓ 33.1%
  203. 203 instructblip-vicuna-7b ✓ 32.7%
  204. 204 Chameleon-30B ✓ 31.8%
  205. 205 Chameleon-7B ✓ 31.1%
  206. 206 SmolVLM2-256M-Video-Instruct ✓ 30.9%
  207. 207 IDEFICS-80B-Instruct ✓ 26.1%
  208. 208 Kosmos-2 ✓ 24.9%
  209. 209 IDEFICS-9B-Instruct ✓ 21.6%

Sources: Labs’ model cards and official MMStar leaderboard

#ModelScoreSettingDateSource
1 Qwen 3.8 Max 85.9% Aug 2026 Lab model card
2 Qwen3.5 397B-A17B 83.8% Feb 2026 Lab model card
3 Qwen 3.6 Plus 83.3% Apr 2026 Lab model card
4 Seed2.0 Pro 83% Feb 2026 Lab model card
5 Qwen3.5-122B-A10B 82.9% Feb 2026 Lab model card
6 Qwen3.5-35B-A3B 81.9% Feb 2026 Lab model card
7 Qwen3.6 27B 81.4% Apr 2026 Lab model card
8 Qwen3.5-27B 81% Feb 2026 Lab model card
9 MiniCPM-V-4.7-35B-A3B 80.7% Oct 2026 Lab model card
10 Seed1.8 79.9% Dec 2025 Lab model card
11 Qwen3.5-9B 79.7% Mar 2026 Lab model card
12 Qwen3.5-Omni-Plus 79.4% Mar 2026 Lab model card
13 Qwen3-VL-235B-A22B 78.7% Sep 2025 Lab model card
14 Qwen3.5-4B 78.3% Mar 2026 Lab model card
15 InternVL3.5-241B-A28B 77.9% Aug 2025 Lab model card
16 Seed1.5-VL 77.8% May 2025 Lab model card
17 STEP3-VL-10B 77.5% Jan 2026 Lab model card
18 GLM-4.6V 75.9% Dec 2025 Lab model card
19 Keye-VL-2.0-30B-A3B 75.9% May 2026 Lab model card
20 ERNIE 5.0 75.5% Feb 2026 Lab model card
21 GLM-4.5V 75.3% Aug 2025 Lab model card
22 Intern-S1 74.9% Jul 2025 Lab model card
23 EXAONE 4.5 33B 74.9% Apr 2026 Lab model card
24 GLM-4.1V-9B-Thinking 72.9% Jul 2025 Lab model card
25 Nemotron Nano 12B v2 VL 71.7% Nov 2025 Lab model card
26 Qwen3.5-2B 71.7% Mar 2026 Lab model card
27 ERNIE-4.5-VL-28B-A3B-Thinking 71.1% Nov 2025 Lab model card
28 LongCat-Flash-Omni 70.9% Oct 2025 Lab model card
29 Qwen2.5-VL-72B 70.8% Jan 2025 Lab model card
30 Kimi-VL-A3B-Thinking-2506 70.4% Jun 2025 Lab model card
31 Qwen3-Omni-30B-A3B 68.5% Sep 2025 Lab model card
32 Phi-4-Reasoning-Vision-15B 64.5% Mar 2026 Lab model card
33 Qwen2.5-Omni-7B 64% Mar 2025 Lab model card
34 LFM2.5-VL-3B 63.3% Aug 2026 Lab model card
35 Qwen3.5-0.8B 58.3% Mar 2026 Lab model card
36 North Micro Vision Instruct 51.8% Aug 2026 Lab model card
37 LFM2-VL-1.6B 49.9% Nov 2025 Lab model card
#ModelScoreSettingDateSource
1 GPT-5 75.7% Aug 2025 Official leaderboard
2 Gemini 2.5 Pro Mar 2025 73.6% Mar 2025 Official leaderboard
3 InternVL3-78B 73.4% Apr 2025 Official leaderboard
4 GPT-5 mini 73.2% Aug 2025 Official leaderboard
5 InternVL3-38B 72.6% Apr 2025 Official leaderboard
6 InternVL2.5-78B-MPO 72.1% Dec 2024 Official leaderboard
7 Qwen2.5-VL-72B 70.5% Jan 2025 Official leaderboard
8 Qwen2.5-VL-32B 70.3% Mar 2025 Official leaderboard
9 InternVL2.5-38B-MPO 70.1% Dec 2024 Official leaderboard
10 Kimi-VL-A3B-Thinking-2506 70% Jun 2025 Official leaderboard
11 GPT-4.1 69.8% Apr 2025 Official leaderboard
12 InternVL2_5-78B 69.5% Dec 2024 Official leaderboard
13 SAIL-VL-1.6-8B 69.5% Apr 2025 Official leaderboard
14 Gemini 2.0 Flash Dec 2024 69.4% Dec 2024 Official leaderboard
15 GPT-4.5 69.3% Feb 2025 Official leaderboard
16 Ovis2-34B 69.2% Feb 2025 Official leaderboard
17 InternVL3-14B 68.9% Apr 2025 Official leaderboard
18 InternVL3-8B 68.7% Apr 2025 Official leaderboard
19 Qwen2-VL-72B-Instruct 68.6% Aug 2024 Official leaderboard
20 InternVL2.5-38B 68.5% Dec 2024 Official leaderboard
21 Gemini 2.0 Pro 68.5% Feb 2025 Official leaderboard
22 GPT-5 nano 68.3% Aug 2025 Official leaderboard
23 InternVL2.5-26B-MPO 67.7% Dec 2024 Official leaderboard
24 InternVL3-9B 67.4% Apr 2025 Official leaderboard
25 Ovis2-16B 67.2% Feb 2025 Official leaderboard
26 InternVL2-Llama3-76B 67.1% Jul 2024 Official leaderboard
27 Gemini 1.5 Pro Sep 2024 67.1% Sep 2024 Official leaderboard
28 InternVL2.5-26B 66.5% Dec 2024 Official leaderboard
29 LLaVA-OneVision 72B 65.8% Aug 2024 Official leaderboard
30 Grok 2 Vision (1212) 65.2% Dec 2024 Official leaderboard
31 InternVL2.5-8B-MPO 65.2% Dec 2024 Official leaderboard
32 Claude 3.5 Sonnet (October 2024) 65.1% Oct 2024 Official leaderboard
33 GPT-4o Nov 2024 65.1% detail-high Nov 2024 Official leaderboard
34 Claude 3.7 Sonnet 65.1% Feb 2025 Official leaderboard
35 InternVL2-40B 64.7% Jul 2024 Official leaderboard
36 GPT-4o Aug 2024 64.7% detail-high Aug 2024 Official leaderboard
37 Ovis2-8B 64.6% Feb 2025 Official leaderboard
38 Gemini 1.5 Flash Sep 2024 64.4% Aug 2024 Official leaderboard
39 Qwen2.5-VL-7B 64.1% Jan 2025 Official leaderboard
40 GPT-4o May 2024 63.9% detail-high May 2024 Official leaderboard
41 NVLM-D-72B 63.7% Sep 2024 Official leaderboard
42 Ovis1.6-Gemma2-27B 63.5% Nov 2024 Official leaderboard
43 Molmo-72B 63.3% Sep 2024 Official leaderboard
44 MiniCPM-o-2_6 63.3% Jan 2025 Official leaderboard
45 InternVL2.5-8B 63.2% Dec 2024 Official leaderboard
46 Ovis1.6-Gemma2-9B 62.9% Sep 2024 Official leaderboard
47 InternVL2-8B-MPO 62.5% Nov 2024 Official leaderboard
48 Claude 3.5 Sonnet 62.2% Jun 2024 Official leaderboard
49 Kimi-VL-A3B-Instruct 62% Apr 2025 Official leaderboard
50 LLaVA-OneVision 7B 61.9% Aug 2024 Official leaderboard
51 DeepSeek-VL2 61.9% Dec 2024 Official leaderboard
52 Ovis2-4B 61.9% Feb 2025 Official leaderboard
53 InternVL2-8B 61.5% Jul 2024 Official leaderboard
54 InternVL3-2B 61.1% Apr 2025 Official leaderboard
55 InternVL2-26B 61% Jul 2024 Official leaderboard
56 InternVL2.5-4B-MPO 61% Dec 2024 Official leaderboard
57 GPT-4.1 mini 60.9% Apr 2025 Official leaderboard
58 Kimi-VL-A3B-Thinking 60.8% Apr 2025 Official leaderboard
59 Qwen2-VL-7B-Instruct 60.7% Aug 2024 Official leaderboard
60 Aria 60.7% Sep 2024 Official leaderboard
61 VITA-1.5 60.2% Dec 2024 Official leaderboard
62 InternLM-XComposer2.5-7B 59.9% Jul 2024 Official leaderboard
63 Gemma 3 27B 59.6% Mar 2025 Official leaderboard
64 Gemini 1.5 Pro May 2024 59.1% May 2024 Official leaderboard
65 Phi-4-multimodal 58.9% Feb 2025 Official leaderboard
66 InternVL2.5-4B 58.7% Dec 2024 Official leaderboard
67 Ovis1.5-Gemma2-9B 58.1% Jul 2024 Official leaderboard
68 MiniCPM-V-2_6 57.5% Aug 2024 Official leaderboard
69 DeepSeek-VL2-Small 57.5% Dec 2024 Official leaderboard
70 Ovis1.5-Llama3-8B 57.3% Jul 2024 Official leaderboard
71 InternVL-Chat-V1-5 57.1% Apr 2024 Official leaderboard
72 LLaVA-OneVision 7B (SI) 56.7% Aug 2024 Official leaderboard
73 Ovis2-2B 56.7% Feb 2025 Official leaderboard
74 SAIL-VL-2B 56.5% Dec 2024 Official leaderboard
75 Qwen2.5-VL-3B 56.3% Jan 2025 Official leaderboard
76 Molmo-7B-D 56.1% Sep 2024 Official leaderboard
77 Gemma 3 12B 56.1% Mar 2025 Official leaderboard
78 GPT-4V 56% detail-high Nov 2023 Official leaderboard
79 Eagle-X5-34B-Chat 55.5% Sep 2024 Official leaderboard
80 InternLM-XComposer2-4KHD-7B 55.3% Apr 2024 Official leaderboard
81 Llama 3.2 90B 55.3% Sep 2024 Official leaderboard
82 VILA1.5-40B 55.2% Apr 2024 Official leaderboard
83 Idefics3-8B-Llama3 55% Aug 2024 Official leaderboard
84 Aquila-VL-2B 54.9% Oct 2024 Official leaderboard
85 InternVL2.5-2B-MPO 54.9% Dec 2024 Official leaderboard
86 GLM-4V-9B 54.8% Jun 2024 Official leaderboard
87 GPT-4o mini 54.8% detail-high Jul 2024 Official leaderboard
88 Granite Vision 3.2 2B 54.6% Feb 2025 Official leaderboard
89 Pixtral 12B 54.5% Sep 2024 Official leaderboard
90 InternVL2.5-2B 54.3% Dec 2024 Official leaderboard
91 Cambrian-1-34B 54.2% Jun 2024 Official leaderboard
92 InternVL2-4B 53.9% Jul 2024 Official leaderboard
93 Mini-InternVL-Chat-4B-V1-5 53.1% May 2024 Official leaderboard
94 InternVL3-1B 52.3% Apr 2025 Official leaderboard
95 Ovis2-1B 52.1% Feb 2025 Official leaderboard
96 Ovis1.6-Llama3.2-3B 52% Oct 2024 Official leaderboard
97 Granite Vision 3.3 2B 51.9% Jun 2025 Official leaderboard
98 MiniCPM-Llama3-V 2.5 51.8% May 2024 Official leaderboard
99 LLaVA-v1.6-34B 51.6% Jan 2024 Official leaderboard
100 InternVL2.5-1B 51.3% Dec 2024 Official leaderboard
101 Cambrian-1-8B 50.7% Jun 2024 Official leaderboard
102 CogVLM2-19B-Chat 50.5% May 2024 Official leaderboard
103 Reka Flash 50.2% Feb 2024 Official leaderboard
104 Molmo-7B-O 50.1% Sep 2024 Official leaderboard
105 InternVL2-2B 49.8% Jul 2024 Official leaderboard
106 Llama 3.2 11B 49.8% Sep 2024 Official leaderboard
107 DeepSeek-VL2-Tiny 49.8% Dec 2024 Official leaderboard
108 Qwen-VL-Max 49.5% Jan 2024 Official leaderboard
109 Idefics2-8B 49.5% Apr 2024 Official leaderboard
110 InternVL2.5-1B-MPO 49.5% Dec 2024 Official leaderboard
111 PaliGemma-3B-mix-448 48.3% May 2024 Official leaderboard
112 Gemma 3 4B 47.9% Mar 2025 Official leaderboard
113 Phi-3-vision-128k-instruct 47.7% May 2024 Official leaderboard
114 Phi-3.5-vision-instruct 47.5% Aug 2024 Official leaderboard
115 Qwen2-VL-2B-Instruct 47.5% Aug 2024 Official leaderboard
116 GPT-4.1 nano 47.5% Apr 2025 Official leaderboard
117 Cambrian-1-13B 47.1% Jun 2024 Official leaderboard
118 Mini-InternVL-Chat-2B-V1-5 46.7% May 2024 Official leaderboard
119 VITA 46.6% Aug 2024 Official leaderboard
120 Emu3-Chat 46.6% Sep 2024 Official leaderboard
121 Janus-Pro-7B 46.5% Jan 2025 Official leaderboard
122 InternLM-XComposer2-VL-1.8B 46.4% Apr 2024 Official leaderboard
123 SmolVLM2-2.2B-Instruct 46% Feb 2025 Official leaderboard
124 Claude 3 Opus 45.7% Feb 2024 Official leaderboard
125 InternVL2-1B 45.6% Jul 2024 Official leaderboard
126 Bunny-Llama-3-8B-V 45.4% Apr 2024 Official leaderboard
127 Falcon2-11B-VLM 45.4% May 2024 Official leaderboard
128 LLaVA-NeXT-Interleave-7B-DPO 45.3% Jun 2024 Official leaderboard
129 LLaVA-NeXT-Interleave-7B 44.5% Jun 2024 Official leaderboard
130 Claude 3 Sonnet 44.2% Feb 2024 Official leaderboard
131 VILA1.5-13B 44.2% Apr 2024 Official leaderboard
132 llama3-llava-next-8b 43.9% Apr 2024 Official leaderboard
133 Eagle-X5-13B 43.7% Aug 2024 Official leaderboard
134 MolmoE-1B 43.1% Sep 2024 Official leaderboard
135 Eagle-X5-7B 41.7% Aug 2024 Official leaderboard
136 SmolVLM-Instruct 41.7% Nov 2024 Official leaderboard
137 Emu2-Chat 40.7% Dec 2023 Official leaderboard
138 Monkey-Chat 40.7% Jan 2024 Official leaderboard
139 VILA1.5-3B 40.6% Apr 2024 Official leaderboard
140 Yi-VL-34B 40.5% Dec 2023 Official leaderboard
141 DeepSeek-VL-7B-Chat 40.5% Mar 2024 Official leaderboard
142 llava-v1.6-vicuna-13b 40.4% Jan 2024 Official leaderboard
143 CogVLM-17B-Chat 39.9% Nov 2023 Official leaderboard
144 DeepSeek-VL-1.3B-Chat 39.9% Mar 2024 Official leaderboard
145 Qwen-VL-Plus 39.7% Jan 2024 Official leaderboard
146 Llama-3-VILA1.5-8B 39.7% Apr 2024 Official leaderboard
147 OmniLMM-12B 39.6% Jan 2024 Official leaderboard
148 MiniCPM-V 2.0 39.1% Apr 2024 Official leaderboard
149 Gemini 1.0 Pro 38.6% Dec 2023 Official leaderboard
150 MiniCPM-V 38.6% Jan 2024 Official leaderboard
151 llava-v1.6-mistral-7b 38.4% Jan 2024 Official leaderboard
152 ShareGPT4V-13B 38.3% Dec 2023 Official leaderboard
153 SmolVLM-500M-Instruct 38.3% Jan 2025 Official leaderboard
154 SmolVLM2-500M-Video-Instruct 38.2% Feb 2025 Official leaderboard
155 Claude 3 Haiku 38.1% Mar 2024 Official leaderboard
156 Yi-VL-6B 37.7% Dec 2023 Official leaderboard
157 LLaVA-OneVision 0.5B 37.7% Aug 2024 Official leaderboard
158 llava-v1.6-vicuna-7b 37.6% Jan 2024 Official leaderboard
159 Janus-1.3B 37.6% Oct 2024 Official leaderboard
160 LLaVA-OneVision 0.5B (SI) 37.5% Aug 2024 Official leaderboard
161 Monkey 37% Nov 2023 Official leaderboard
162 ShareGPT4V-7B 35.7% Nov 2023 Official leaderboard
163 mPLUG-Owl2 34.8% Nov 2023 Official leaderboard
164 Qwen-VL-Chat 34.5% Aug 2023 Official leaderboard
165 SmolVLM-256M-Instruct 34.5% Jan 2025 Official leaderboard
166 LLaVA-v1.5-13B 34.3% Oct 2023 Official leaderboard
167 llava-v1.5-7b 33.1% Oct 2023 Official leaderboard
168 instructblip-vicuna-7b 32.7% May 2023 Official leaderboard
169 Chameleon-30B 31.8% Jun 2024 Official leaderboard
170 Chameleon-7B 31.1% Jun 2024 Official leaderboard
171 SmolVLM2-256M-Video-Instruct 30.9% Feb 2025 Official leaderboard
172 IDEFICS-80B-Instruct 26.1% Jul 2023 Official leaderboard
173 Kosmos-2 24.9% Jun 2023 Official leaderboard
174 IDEFICS-9B-Instruct 21.6% Jul 2023 Official leaderboard
#ModelScoreSettingDateSource
1 Qwen 3.8 Max 85.9% Aug 2026 Lab model card
2 Qwen3.5 397B-A17B 83.8% Feb 2026 Lab model card
3 Qwen 3.6 Plus 83.3% Apr 2026 Lab model card
4 Seed2.0 Pro 83% Feb 2026 Lab model card
5 Qwen3.5-122B-A10B 82.9% Feb 2026 Lab model card
6 Qwen3.5-35B-A3B 81.9% Feb 2026 Lab model card
7 Qwen3.6 27B 81.4% Apr 2026 Lab model card
8 Qwen3.5-27B 81% Feb 2026 Lab model card
9 MiniCPM-V-4.7-35B-A3B 80.7% Oct 2026 Lab model card
10 Seed1.8 79.9% Dec 2025 Lab model card
11 Qwen3.5-9B 79.7% Mar 2026 Lab model card
12 Qwen3.5-Omni-Plus 79.4% Mar 2026 Lab model card
13 Qwen3-VL-235B-A22B 78.7% Sep 2025 Lab model card
14 Qwen3.5-4B 78.3% Mar 2026 Lab model card
15 InternVL3.5-241B-A28B 77.9% Aug 2025 Lab model card
16 Seed1.5-VL 77.8% May 2025 Lab model card
17 STEP3-VL-10B 77.5% Jan 2026 Lab model card
18 GLM-4.6V 75.9% Dec 2025 Lab model card
19 Keye-VL-2.0-30B-A3B 75.9% May 2026 Lab model card
20 GPT-5 75.7% Aug 2025 Official leaderboard
21 ERNIE 5.0 75.5% Feb 2026 Lab model card
22 GLM-4.5V 75.3% Aug 2025 Lab model card
23 Intern-S1 74.9% Jul 2025 Lab model card
24 EXAONE 4.5 33B 74.9% Apr 2026 Lab model card
25 Gemini 2.5 Pro Mar 2025 73.6% Mar 2025 Official leaderboard
26 InternVL3-78B 73.4% Apr 2025 Official leaderboard
27 GPT-5 mini 73.2% Aug 2025 Official leaderboard
28 GLM-4.1V-9B-Thinking 72.9% Jul 2025 Lab model card
29 InternVL3-38B 72.6% Apr 2025 Official leaderboard
30 InternVL2.5-78B-MPO 72.1% Dec 2024 Official leaderboard
31 Nemotron Nano 12B v2 VL 71.7% Nov 2025 Lab model card
32 Qwen3.5-2B 71.7% Mar 2026 Lab model card
33 ERNIE-4.5-VL-28B-A3B-Thinking 71.1% Nov 2025 Lab model card
34 LongCat-Flash-Omni 70.9% Oct 2025 Lab model card
35 Qwen2.5-VL-72B 70.8% Jan 2025 Lab model card
36 Kimi-VL-A3B-Thinking-2506 70.4% Jun 2025 Lab model card
37 Qwen2.5-VL-32B 70.3% Mar 2025 Official leaderboard
38 InternVL2.5-38B-MPO 70.1% Dec 2024 Official leaderboard
39 GPT-4.1 69.8% Apr 2025 Official leaderboard
40 InternVL2_5-78B 69.5% Dec 2024 Official leaderboard
41 SAIL-VL-1.6-8B 69.5% Apr 2025 Official leaderboard
42 Gemini 2.0 Flash Dec 2024 69.4% Dec 2024 Official leaderboard
43 GPT-4.5 69.3% Feb 2025 Official leaderboard
44 Ovis2-34B 69.2% Feb 2025 Official leaderboard
45 InternVL3-14B 68.9% Apr 2025 Official leaderboard
46 InternVL3-8B 68.7% Apr 2025 Official leaderboard
47 Qwen2-VL-72B-Instruct 68.6% Aug 2024 Official leaderboard
48 InternVL2.5-38B 68.5% Dec 2024 Official leaderboard
49 Gemini 2.0 Pro 68.5% Feb 2025 Official leaderboard
50 Qwen3-Omni-30B-A3B 68.5% Sep 2025 Lab model card
51 GPT-5 nano 68.3% Aug 2025 Official leaderboard
52 InternVL2.5-26B-MPO 67.7% Dec 2024 Official leaderboard
53 InternVL3-9B 67.4% Apr 2025 Official leaderboard
54 Ovis2-16B 67.2% Feb 2025 Official leaderboard
55 InternVL2-Llama3-76B 67.1% Jul 2024 Official leaderboard
56 Gemini 1.5 Pro Sep 2024 67.1% Sep 2024 Official leaderboard
57 InternVL2.5-26B 66.5% Dec 2024 Official leaderboard
58 LLaVA-OneVision 72B 65.8% Aug 2024 Official leaderboard
59 Grok 2 Vision (1212) 65.2% Dec 2024 Official leaderboard
60 InternVL2.5-8B-MPO 65.2% Dec 2024 Official leaderboard
61 Claude 3.5 Sonnet (October 2024) 65.1% Oct 2024 Official leaderboard
62 GPT-4o Nov 2024 65.1% detail-high Nov 2024 Official leaderboard
63 Claude 3.7 Sonnet 65.1% Feb 2025 Official leaderboard
64 InternVL2-40B 64.7% Jul 2024 Official leaderboard
65 GPT-4o Aug 2024 64.7% detail-high Aug 2024 Official leaderboard
66 Ovis2-8B 64.6% Feb 2025 Official leaderboard
67 Phi-4-Reasoning-Vision-15B 64.5% Mar 2026 Lab model card
68 Gemini 1.5 Flash Sep 2024 64.4% Aug 2024 Official leaderboard
69 Qwen2.5-VL-7B 64.1% Jan 2025 Official leaderboard
70 Qwen2.5-Omni-7B 64% Mar 2025 Lab model card
71 GPT-4o May 2024 63.9% detail-high May 2024 Official leaderboard
72 NVLM-D-72B 63.7% Sep 2024 Official leaderboard
73 Ovis1.6-Gemma2-27B 63.5% Nov 2024 Official leaderboard
74 Molmo-72B 63.3% Sep 2024 Official leaderboard
75 MiniCPM-o-2_6 63.3% Jan 2025 Official leaderboard
76 LFM2.5-VL-3B 63.3% Aug 2026 Lab model card
77 InternVL2.5-8B 63.2% Dec 2024 Official leaderboard
78 Ovis1.6-Gemma2-9B 62.9% Sep 2024 Official leaderboard
79 InternVL2-8B-MPO 62.5% Nov 2024 Official leaderboard
80 Claude 3.5 Sonnet 62.2% Jun 2024 Official leaderboard
81 Kimi-VL-A3B-Instruct 62% Apr 2025 Official leaderboard
82 LLaVA-OneVision 7B 61.9% Aug 2024 Official leaderboard
83 DeepSeek-VL2 61.9% Dec 2024 Official leaderboard
84 Ovis2-4B 61.9% Feb 2025 Official leaderboard
85 InternVL2-8B 61.5% Jul 2024 Official leaderboard
86 InternVL3-2B 61.1% Apr 2025 Official leaderboard
87 InternVL2-26B 61% Jul 2024 Official leaderboard
88 InternVL2.5-4B-MPO 61% Dec 2024 Official leaderboard
89 GPT-4.1 mini 60.9% Apr 2025 Official leaderboard
90 Kimi-VL-A3B-Thinking 60.8% Apr 2025 Official leaderboard
91 Qwen2-VL-7B-Instruct 60.7% Aug 2024 Official leaderboard
92 Aria 60.7% Sep 2024 Official leaderboard
93 VITA-1.5 60.2% Dec 2024 Official leaderboard
94 InternLM-XComposer2.5-7B 59.9% Jul 2024 Official leaderboard
95 Gemma 3 27B 59.6% Mar 2025 Official leaderboard
96 Gemini 1.5 Pro May 2024 59.1% May 2024 Official leaderboard
97 Phi-4-multimodal 58.9% Feb 2025 Official leaderboard
98 InternVL2.5-4B 58.7% Dec 2024 Official leaderboard
99 Qwen3.5-0.8B 58.3% Mar 2026 Lab model card
100 Ovis1.5-Gemma2-9B 58.1% Jul 2024 Official leaderboard
101 MiniCPM-V-2_6 57.5% Aug 2024 Official leaderboard
102 DeepSeek-VL2-Small 57.5% Dec 2024 Official leaderboard
103 Ovis1.5-Llama3-8B 57.3% Jul 2024 Official leaderboard
104 InternVL-Chat-V1-5 57.1% Apr 2024 Official leaderboard
105 LLaVA-OneVision 7B (SI) 56.7% Aug 2024 Official leaderboard
106 Ovis2-2B 56.7% Feb 2025 Official leaderboard
107 SAIL-VL-2B 56.5% Dec 2024 Official leaderboard
108 Qwen2.5-VL-3B 56.3% Jan 2025 Official leaderboard
109 Molmo-7B-D 56.1% Sep 2024 Official leaderboard
110 Gemma 3 12B 56.1% Mar 2025 Official leaderboard
111 GPT-4V 56% detail-high Nov 2023 Official leaderboard
112 Eagle-X5-34B-Chat 55.5% Sep 2024 Official leaderboard
113 InternLM-XComposer2-4KHD-7B 55.3% Apr 2024 Official leaderboard
114 Llama 3.2 90B 55.3% Sep 2024 Official leaderboard
115 VILA1.5-40B 55.2% Apr 2024 Official leaderboard
116 Idefics3-8B-Llama3 55% Aug 2024 Official leaderboard
117 Aquila-VL-2B 54.9% Oct 2024 Official leaderboard
118 InternVL2.5-2B-MPO 54.9% Dec 2024 Official leaderboard
119 GLM-4V-9B 54.8% Jun 2024 Official leaderboard
120 GPT-4o mini 54.8% detail-high Jul 2024 Official leaderboard
121 Granite Vision 3.2 2B 54.6% Feb 2025 Official leaderboard
122 Pixtral 12B 54.5% Sep 2024 Official leaderboard
123 InternVL2.5-2B 54.3% Dec 2024 Official leaderboard
124 Cambrian-1-34B 54.2% Jun 2024 Official leaderboard
125 InternVL2-4B 53.9% Jul 2024 Official leaderboard
126 Mini-InternVL-Chat-4B-V1-5 53.1% May 2024 Official leaderboard
127 InternVL3-1B 52.3% Apr 2025 Official leaderboard
128 Ovis2-1B 52.1% Feb 2025 Official leaderboard
129 Ovis1.6-Llama3.2-3B 52% Oct 2024 Official leaderboard
130 Granite Vision 3.3 2B 51.9% Jun 2025 Official leaderboard
131 MiniCPM-Llama3-V 2.5 51.8% May 2024 Official leaderboard
132 North Micro Vision Instruct 51.8% Aug 2026 Lab model card
133 LLaVA-v1.6-34B 51.6% Jan 2024 Official leaderboard
134 InternVL2.5-1B 51.3% Dec 2024 Official leaderboard
135 Cambrian-1-8B 50.7% Jun 2024 Official leaderboard
136 CogVLM2-19B-Chat 50.5% May 2024 Official leaderboard
137 Reka Flash 50.2% Feb 2024 Official leaderboard
138 Molmo-7B-O 50.1% Sep 2024 Official leaderboard
139 LFM2-VL-1.6B 49.9% Nov 2025 Lab model card
140 InternVL2-2B 49.8% Jul 2024 Official leaderboard
141 Llama 3.2 11B 49.8% Sep 2024 Official leaderboard
142 DeepSeek-VL2-Tiny 49.8% Dec 2024 Official leaderboard
143 Qwen-VL-Max 49.5% Jan 2024 Official leaderboard
144 Idefics2-8B 49.5% Apr 2024 Official leaderboard
145 InternVL2.5-1B-MPO 49.5% Dec 2024 Official leaderboard
146 PaliGemma-3B-mix-448 48.3% May 2024 Official leaderboard
147 Gemma 3 4B 47.9% Mar 2025 Official leaderboard
148 Phi-3-vision-128k-instruct 47.7% May 2024 Official leaderboard
149 Phi-3.5-vision-instruct 47.5% Aug 2024 Official leaderboard
150 Qwen2-VL-2B-Instruct 47.5% Aug 2024 Official leaderboard
151 GPT-4.1 nano 47.5% Apr 2025 Official leaderboard
152 Cambrian-1-13B 47.1% Jun 2024 Official leaderboard
153 Mini-InternVL-Chat-2B-V1-5 46.7% May 2024 Official leaderboard
154 VITA 46.6% Aug 2024 Official leaderboard
155 Emu3-Chat 46.6% Sep 2024 Official leaderboard
156 Janus-Pro-7B 46.5% Jan 2025 Official leaderboard
157 InternLM-XComposer2-VL-1.8B 46.4% Apr 2024 Official leaderboard
158 SmolVLM2-2.2B-Instruct 46% Feb 2025 Official leaderboard
159 Claude 3 Opus 45.7% Feb 2024 Official leaderboard
160 InternVL2-1B 45.6% Jul 2024 Official leaderboard
161 Bunny-Llama-3-8B-V 45.4% Apr 2024 Official leaderboard
162 Falcon2-11B-VLM 45.4% May 2024 Official leaderboard
163 LLaVA-NeXT-Interleave-7B-DPO 45.3% Jun 2024 Official leaderboard
164 LLaVA-NeXT-Interleave-7B 44.5% Jun 2024 Official leaderboard
165 Claude 3 Sonnet 44.2% Feb 2024 Official leaderboard
166 VILA1.5-13B 44.2% Apr 2024 Official leaderboard
167 llama3-llava-next-8b 43.9% Apr 2024 Official leaderboard
168 Eagle-X5-13B 43.7% Aug 2024 Official leaderboard
169 MolmoE-1B 43.1% Sep 2024 Official leaderboard
170 Eagle-X5-7B 41.7% Aug 2024 Official leaderboard
171 SmolVLM-Instruct 41.7% Nov 2024 Official leaderboard
172 Emu2-Chat 40.7% Dec 2023 Official leaderboard
173 Monkey-Chat 40.7% Jan 2024 Official leaderboard
174 VILA1.5-3B 40.6% Apr 2024 Official leaderboard
175 Yi-VL-34B 40.5% Dec 2023 Official leaderboard
176 DeepSeek-VL-7B-Chat 40.5% Mar 2024 Official leaderboard
177 llava-v1.6-vicuna-13b 40.4% Jan 2024 Official leaderboard
178 CogVLM-17B-Chat 39.9% Nov 2023 Official leaderboard
179 DeepSeek-VL-1.3B-Chat 39.9% Mar 2024 Official leaderboard
180 Qwen-VL-Plus 39.7% Jan 2024 Official leaderboard
181 Llama-3-VILA1.5-8B 39.7% Apr 2024 Official leaderboard
182 OmniLMM-12B 39.6% Jan 2024 Official leaderboard
183 MiniCPM-V 2.0 39.1% Apr 2024 Official leaderboard
184 Gemini 1.0 Pro 38.6% Dec 2023 Official leaderboard
185 MiniCPM-V 38.6% Jan 2024 Official leaderboard
186 llava-v1.6-mistral-7b 38.4% Jan 2024 Official leaderboard
187 ShareGPT4V-13B 38.3% Dec 2023 Official leaderboard
188 SmolVLM-500M-Instruct 38.3% Jan 2025 Official leaderboard
189 SmolVLM2-500M-Video-Instruct 38.2% Feb 2025 Official leaderboard
190 Claude 3 Haiku 38.1% Mar 2024 Official leaderboard
191 Yi-VL-6B 37.7% Dec 2023 Official leaderboard
192 LLaVA-OneVision 0.5B 37.7% Aug 2024 Official leaderboard
193 llava-v1.6-vicuna-7b 37.6% Jan 2024 Official leaderboard
194 Janus-1.3B 37.6% Oct 2024 Official leaderboard
195 LLaVA-OneVision 0.5B (SI) 37.5% Aug 2024 Official leaderboard
196 Monkey 37% Nov 2023 Official leaderboard
197 ShareGPT4V-7B 35.7% Nov 2023 Official leaderboard
198 mPLUG-Owl2 34.8% Nov 2023 Official leaderboard
199 Qwen-VL-Chat 34.5% Aug 2023 Official leaderboard
200 SmolVLM-256M-Instruct 34.5% Jan 2025 Official leaderboard
201 LLaVA-v1.5-13B 34.3% Oct 2023 Official leaderboard
202 llava-v1.5-7b 33.1% Oct 2023 Official leaderboard
203 instructblip-vicuna-7b 32.7% May 2023 Official leaderboard
204 Chameleon-30B 31.8% Jun 2024 Official leaderboard
205 Chameleon-7B 31.1% Jun 2024 Official leaderboard
206 SmolVLM2-256M-Video-Instruct 30.9% Feb 2025 Official leaderboard
207 IDEFICS-80B-Instruct 26.1% Jul 2023 Official leaderboard
208 Kosmos-2 24.9% Jun 2023 Official leaderboard
209 IDEFICS-9B-Instruct 21.6% Jul 2023 Official leaderboard

The benchmark's own leaderboard has 277 entries, 174 of them models we track. Switch the scores above to rank them together.

8 models have held number one in labs’ own numbers since its release in March 2024 10 models have held number one across all independent runs since its release in March 2024 16 models have held number one with claims included since its release in March 2024

Lab claims count from the day the card came out, independent runs from the model’s release. We draw one row each for the benchmark’s own leaderboard and what labs claim in their cards.

OpenAI Other labs Shanghai AI Lab Google DeepMind Qwen ByteDance Seed
Own leaderboard Not tested since Aug 2025
GPT-5 75.7%
Lab claims Passed 85% in Aug 2026, by Qwen 3.8 Max
Qwen 3.8 Max 85.9%

Sources: Official MMStar leaderboard and labs’ model cards

Qwen’s own number for Qwen 3.8 Max passed 85% in August 2026

That was 29 months after its release. By the same age, 40% of the benchmarks from 2024 had passed 85%. All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
On the benchmark’s own leaderboard Reported by the lab in its model card Best so far
0% 20% 40% 60% 80% 100% 2023 2024 2025 2026 Near the ceiling (85%) Qwen 3.8 Max Seed1.8 Seed1.5-VL Gemini 2.5 Pro Qwen2-VL-72B-Instruct GPT-4o InternVL-Chat-V1-5 0% 20% 40% 60% 80% 100% 2023 2024 2025 2026 Near the ceiling (85%) Qwen 3.8 Max Qwen3-VL-235B-A22B

Sources: Official MMStar leaderboard and labs’ model cards

4 labs put it in a launch post in the last six months

Launch posts from every lab that has listed it, one tick per post. Tall colored ticks include it, and short gray ones leave it out.

Qwen
ByteDance Seed
Moonshot AI
Z.ai
Shanghai AI Lab
Liquid AI
Meituan
Baidu
StepFun
Microsoft
LG AI Research
Kuaishou
Cohere
OpenBMB (ModelBest)
DateLaunch postScoreAs the post writes it
Oct 7, 2026 MiniCPM-V-4.7-35B-A3B 80.7 MMStar
Aug 12, 2026 LFM2.5-VL-3B 63.3 MMStar
Aug 12, 2026 North Micro Vision Instruct 51.8 MMStar
Aug 3, 2026 Qwen 3.8 Max 85.9 MMStar
May 25, 2026 Keye-VL-2.0-30B-A3B 75.9 MMStar
Apr 22, 2026 Qwen3.6 27B 81.4 MMStar
Apr 9, 2026 EXAONE 4.5 33B 74.9 MMStar
Apr 2, 2026 Qwen 3.6 Plus 83.3 MMStar
Mar 30, 2026 Qwen3.5-Omni-Plus 79.4 MMStar
Mar 4, 2026 Phi-4-Reasoning-Vision-15B 64.5 MMStar (default mode)
Mar 2, 2026 Qwen3.5-0.8B 58.3 MMStar 58.3/55.9 (with CI / without CI)
Mar 2, 2026 Qwen3.5-2B 71.7 MMStar 71.7/68.0 (with CI / without CI)
Mar 2, 2026 Qwen3.5-4B 78.3 MMStar
Mar 2, 2026 Qwen3.5-9B 79.7 MMStar
Feb 24, 2026 Qwen3.5-35B-A3B 81.9 MMStar
Feb 24, 2026 Qwen3.5-27B 81 MMStar
Feb 24, 2026 Qwen3.5-122B-A10B 82.9 MMStar
Feb 16, 2026 Qwen3.5 397B-A17B 83.8 MMStar
Feb 6, 2026 ERNIE 5.0 75.5 MMStar
Jan 13, 2026 STEP3-VL-10B 77.5 MMStar
Dec 18, 2025 Seed1.8 79.9 MMStar
Dec 8, 2025 GLM-4.6V 75.9 MMStar
Nov 11, 2025 ERNIE-4.5-VL-28B-A3B-Thinking 71.1 MMStar
Oct 31, 2025 LongCat-Flash-Omni 70.9 MMStar
Sep 23, 2025 Qwen3-VL-235B-A22B 78.7 MMStar
Sep 22, 2025 Qwen3-Omni-30B-A3B 68.5 MMStar
Aug 25, 2025 InternVL3.5-241B-A28B 77.9 MMStar
Aug 12, 2025 LFM2-VL-1.6B 49.5 MMStar
Aug 11, 2025 GLM-4.5V 75.3 MMStar
Jul 24, 2025 Intern-S1 74.9 MMStar
Jul 1, 2025 GLM-4.1V-9B-Thinking 72.9 MMStar
Jun 21, 2025 Kimi-VL-A3B-Thinking-2506 70.4 MMStar (Acc)
May 12, 2025 Seed1.5-VL 77.8 MMStar
Mar 27, 2025 Qwen2.5-Omni-7B 64 MMStar
Jan 28, 2025 Qwen2.5-VL-72B 70.8 MMStar

22 system cards report it, from 13 labs

Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.

DateDocumentScoreAs the document writes itSection
Jun 9, 2026 Keye-VL-2.0-30B-A3B technical report 75.9 MMStar (Figure 5, printed value) Capabilities
Apr 17, 2026 Qwen3.5-Omni-Plus technical report 79.4 MMStar (Table 6) Capabilities
Apr 9, 2026 EXAONE 4.5 33B technical report 74.9 MMStar (Table 2, Reasoning mode) Capabilities
Mar 21, 2026 Seed1.8 model card 79.9 MMStar (Table 2) Capabilities
Mar 4, 2026 Phi-4-Reasoning-Vision-15B technical report 64.5 MMStar, default mode, 63.9 force nothink and force thinking (Tables 5 and 6) Capabilities
Feb 14, 2026 Seed2.0 Pro model card 83 MMStar (Table 8) Capabilities
Feb 4, 2026 ERNIE 5.0 technical report 75.5 MMStar (Table 3) Capabilities
Jan 14, 2026 STEP3-VL-10B technical report 77.5 MMStar, SeRe (Tables 1 and 3). PaCoRe 77.64 Capabilities
Jan 1, 2026 GLM-4.6V technical report 75.9 MMStar, GLM-4.6V thinking (Table 2) Capabilities
Nov 28, 2025 LFM2-VL-1.6B technical report 49.9 MMStar, LFM2-VL-1.6B (Table 8) Capabilities
Nov 26, 2025 Qwen3-VL-235B-A22B technical report 78.7 MMStar (Table 2) Capabilities
Nov 6, 2025 Nemotron Nano 12B v2 VL technical report 71.7 MMStar, reasoning on (Table 3). Reasoning off 65.9 Capabilities
Oct 31, 2025 LongCat-Flash-Omni technical report 70.9 MMStar (Table 5) Capabilities
Sep 22, 2025 Qwen3-Omni-30B-A3B technical report 68.5 MMStar, Instruct (Table 9). Thinking 74.9 (Table 10) Capabilities
Aug 25, 2025 InternVL3.5-241B-A28B technical report 77.9 MMStar (Table 2, Table 6) Capabilities
Aug 21, 2025 Intern-S1 technical report 74.9 MMStar (Table 2) Capabilities
Aug 13, 2025 GLM-4.5V technical report 75.3 MMStar, GLM-4.5V thinking (Table 2) Capabilities
Jul 1, 2025 GLM-4.1V-9B-Thinking technical report 72.9 MMStar, GLM-4.1V-9B-Thinking thinking (Table 2) Capabilities
Jun 23, 2025 Kimi-VL-A3B-Thinking-2506 technical report 70.4 MMStar (Acc) (Table 5) Capabilities
May 11, 2025 Seed1.5-VL technical report 77.8 MMStar, thinking (Table 6). Non-thinking 76.2 Capabilities
Mar 26, 2025 Qwen2.5-Omni-7B technical report 64 MMStar (Table 5) Capabilities
Feb 19, 2025 Qwen2.5-VL-72B technical report 70.8 MMStar (Table 3) Capabilities

The lab's number comes out ahead in 2 of 2 pairs that differ

Each score a lab reported for its own model, beside the same model's best score on the leaderboard. Gaps mostly come from different harnesses, effort levels and task subsets. Every benchmark.

Model Lab says Board says Gap
Kimi-VL-A3B-Thinking-2506 70.4% MMStar (Acc) 70% Leaderboard +0.4 pts
Qwen2.5-VL-72B 70.8% MMStar 70.5% Leaderboard +0.3 pts

More in Vision

All 636 in Vision
  • MMMU In use 11.5K college-level multimodal questions across 30 subjects. Multiple-choice and open accuracy. Qwen 3.6 Plus 86%
  • MMMU-Pro In use Harder MMMU: college-level multimodal questions with more options and vision-only inputs to remove text shortcuts. GPT-5.6 Sol 84.6%
  • CharXiv In use Chart understanding on real arXiv figures. Reasoning questions require synthesizing information from complex charts. Claude Fable 5 93.5%
  • MathVista In use Math reasoning in visual contexts (charts, geometry, puzzles) from 31 datasets. Answer accuracy. Seed2.1 Pro 90.7%
  • MATH-Vision (MathVision) In use 3,040 competition math problems with visual context. Answer accuracy. Qwen 3.8 Max 97.7%
  • ChartQA In use Models answer human-written and generated questions about chart images that need visual and arithmetic reasoning. Scored by relaxed accuracy with a 5 percent numeric tolerance. Command A Vision 90.9%

Data audit

All 10 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 3 pass 2 notes
  • Pass Independent runs Scores from the official leaderboard (277 entries), launch posts (35 scores) and system cards (22 scores).
  • Pass Models scored 209 models we track, plus 103 other entries on its leaderboard.
  • Pass Newest score The newest is from October 2026, for MiniCPM-V-4.7-35B-A3B.
  • Note Frontier models None of the top 10 has a score, which is fine for a retired test. The strongest tested is Qwen 3.8 Max (#30).
  • Note Headroom Saturated. The best is 85.9%, past 85% since August 2026.
Leaderboard import 3 pass
  • Pass Importer We read the board on rank.opencompass.org.cn: 180 rows for models we track and 103 other entries at the last read.
  • Pass Last read November 18th, 2026.
  • Pass Read worked The last read worked.
Consistency 2 pass
  • Pass Lab claims Both pairs agree within 10 points.
  • Pass Known issues No published audit found problems.

How we check