Benchmark Almanac

MathVista

Seed2.1 Pro 90.7% leads since June 2026
Doubao-1.5-pro 79.5% leads since January 2025
Seed2.1 Pro 90.7% leads since June 2026
In use

The official leaderboard hasn’t scored a model released after Llama 4 Maverick (April 2025). Labs report their own numbers for 47 newer models, such as Seed2.1 Pro at 90.7%.

Math reasoning in visual contexts (charts, geometry, puzzles) from 31 datasets. Answer accuracy.

Near-saturated.

Made by
UCLA + Microsoft
Released
October 2023
Announced on X
11K views, 94 likes
Kind
Visual math and diagrams
Level
School
Scored
Checked automatically
Answers
Short answer
Tasks
All public
Measures
Capability
Subject
Mathematics
Language
English
In and out
Text, Images → Text
Interaction
Single turn
Refresh
One fixed set
Run by
Anyone
In views
Vision
Strongest model tested
Kimi K2.5 #76 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 Language Models (LLMs) and Large Multimodal Models (LMMs) exhibit impressive problem-solving skills in many tasks and domains, but their ability in mathematical reasoning in visual contexts has not been systematically studied. To bridge this gap, we present MathVista, a benchmark designed to combine challenges from diverse mathematical and visual tasks. It consists of 6,141 examples, derived from 28 existing multimodal datasets involving mathematics and 3 newly created datasets (i.e., IQTest, FunctionQA, and PaperQA). Completing these tasks requires fine-grained, deep visual understanding and compositional reasoning, which all state-of-the-art foundation models find challenging. With MathVista, we have conducted a comprehensive, quantitative evaluation of 12 prominent foundation models. The best-performing GPT-4V model achieves an overall accuracy of 49.9%, substantially outperforming Bard, the second-best performer, by 15.1%. Our in-depth analysis reveals that the superiority of GPT-4V is mainly attributed to its enhanced visual perception and mathematical reasoning. However, GPT-4V still falls short of human performance by 10.4%, as it often struggles to understand complex figures and perform rigorous reasoning. This significant gap underscores the critical role that MathVista will play in the development of general-purpose AI agents capable of tackling mathematically intensive and visually rich real-world tasks. We further explore the new ability of self-verification, the application of self-consistency, and the interactive chatbot capabilities of GPT-4V, highlighting its promising potential for future research. The project is available at https://mathvista.github.io/.

Abstract of MathVista: Evaluating Mathematical Reasoning of Foundation Models in Visual Contexts (arXiv:2310.02255, CC0)
71
Models scored in lab cards
65
Launch posts list it, from 21 labs
42
System cards report it, from 17 labs

Seed2.1 Pro leads with 90.7%

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.

Doubao-1.5-pro leads with 79.5% 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.

Seed2.1 Pro leads with 90.7%, 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 Seed2.1 Pro 90.7%
  2. 2 Qwen3.5 397B-A17B 90.3%
  3. 3 Kimi K2.5 90.1%
  4. 4 Seed2.0 Pro 89.8%
  5. 5 MiniCPM-V-4.7-35B-A3B 88%
  6. 6 Qwen3.5-27B 87.8%
  7. 7 Seed1.8 87.7%
  8. 8 Qwen3.5-122B-A10B 87.4%
  9. 9 Qwen3.6 27B 87.4%
  10. 10 o3 86.8%
  11. 11 Qwen 3.6 35B-A3B 86.4%
  12. 12 Qwen3.5-35B-A3B 86.2%
  13. 13 Qwen3.5-Omni-Plus 86.1%
  14. 14 Qwen3-VL-235B-A22B 85.8%
  15. 15 Qwen3.5-9B 85.7%
  16. 16 Seed1.5-VL 85.6%
  17. 17 GLM-4.6V 85.2%
  18. 18 Qwen3.5-4B 85.1%
  19. 19 EXAONE 4.5 33B 85%
  20. 20 ERNIE 5.0 84.8%
  21. 21 GLM-4.5V 84.6%
  22. 22 o4-mini 84.3%
  23. 23 STEP3-VL-10B 84%
  24. 24 Keye-VL-2.0-30B-A3B 82.9%
  25. 25 Nemotron 3 Nano Omni 30B-A3B 82.8%
  26. 26 InternVL3.5-241B-A28B 82.7%
  27. 27 ERNIE-4.5-VL-28B-A3B-Thinking 82.5%
  28. 28 Intern-S1 81.5%
  29. 29 GLM-4.1V-9B-Thinking 80.7%
  30. 30 Command A+ 80.6%
  31. 31 Kimi-VL-A3B-Thinking-2506 80.1%
  32. 32 LongCat-Flash-Omni 77.9%
  33. 33 Nemotron Nano 12B v2 VL 76.9%
  34. 34 Qwen3.5-2B 76.7%
  35. 35 Qwen3-Omni-30B-A3B 75.9%
  36. 36 Phi-4-Reasoning-Vision-15B 75.2%
  37. 37 Kimi k1.5 74.9%
  38. 38 Qwen2.5-VL-72B 74.8%
  39. 39 Llama 4 Maverick 73.7%
  40. 40 Command A Vision 73.5%
  41. 41 GPT-4.1 mini 73.1%
  42. 42 BAGEL-7B-MoT 73.1%
  43. 43 GPT-4.1 72.2%
  44. 44 Kimi-VL-A3B-Thinking 71.3%
  45. 45 o1 71%
  46. 46 Claude 3.5 Sonnet (October 2024) 70.7%
  47. 47 Llama 4 Scout 70.7%
  48. 48 Magistral Medium 1.0 70.1%
  49. 49 Grok-2 Aug 2024 69%
  50. 50 Mistral Small 3.1 68.9%
  51. 51 LFM2.5-VL-3B 68.5%
  52. 52 Qwen2.5-Omni-7B 67.9%
  53. 53 Claude 3.5 Sonnet 67.7%
  54. 54 Gemma 3 27B 67.6%
  55. 55 Mistral Small 3.2 67.1%
  56. 56 Magistral Small 1.0 66.8%
  57. 57 Gemini 1.5 Pro May 2024 63.9%
  58. 58 GPT-4o May 2024 63.8%
  59. 59 Gemma 3 12B 62.9%
  60. 60 Phi-4-multimodal 62.4%
  61. 61 Qwen3.5-0.8B 62.2%
  62. 62 GPT-4o mini 56.7%
  63. 63 GPT-4.1 nano 56.2%
  64. 64 Gemini 1.0 Ultra 53%
  65. 65 Gemini 1.5 Pro Feb 2024 52.1%
  66. 66 LFM2-VL-1.6B 51.7%
  67. 67 Claude 3 Opus 50.5%
  68. 68 Gemma 3 4B 50%
  69. 69 Claude 3 Sonnet 47.9%
  70. 70 Claude 3 Haiku 46.4%
  71. 71 Phi-3.5-vision-instruct 43.9%

Source: Labs’ model cards

Source: Official MathVista leaderboard

100 models. Scroll sideways for the rest.

  1. 90.7 Seed2.1 Pro
  2. 90.3 Qwen3.5 397B-A17B
  3. 90.1 Kimi K2.5
  4. 89.8 Seed2.0 Pro
  5. 88.0 MiniCPM-V-4.7-35B-A3B
  6. 87.8 Qwen3.5-27B
  7. 87.7 Seed1.8
  8. 87.4 Qwen3.5-122B-A10B
  9. 87.4 Qwen3.6 27B
  10. 86.8 o3
  11. 86.4 Qwen 3.6 35B-A3B
  12. 86.2 Qwen3.5-35B-A3B
  13. 86.1 Qwen3.5-Omni-Plus
  14. 85.8 Qwen3-VL-235B-A22B
  15. 85.7 Qwen3.5-9B
  16. 85.6 Seed1.5-VL
  17. 85.2 GLM-4.6V
  18. 85.1 Qwen3.5-4B
  19. 85.0 EXAONE 4.5 33B
  20. 84.8 ERNIE 5.0
  21. 84.6 GLM-4.5V
  22. 84.3 o4-mini
  23. 84.0 STEP3-VL-10B
  24. 82.9 Keye-VL-2.0-30B-A3B
  25. 82.8 Nemotron 3 Nano Omni 30B-A3B
  26. 82.7 InternVL3.5-241B-A28B
  27. 82.5 ERNIE-4.5-VL-28B-A3B-Thinking
  28. 81.5 Intern-S1
  29. 80.7 GLM-4.1V-9B-Thinking
  30. 80.6 Command A+
  31. 80.1 Kimi-VL-A3B-Thinking-2506
  32. 79.5 Doubao-1.5-pro
  33. 77.9 LongCat-Flash-Omni
  34. 77.1 Ovis2-34B
  35. 76.9 Nemotron Nano 12B v2 VL
  36. 76.7 Qwen3.5-2B
  37. 75.9 Qwen3-Omni-30B-A3B
  38. 75.2 Phi-4-Reasoning-Vision-15B
  39. 74.9 Kimi k1.5
  40. 74.8 Qwen2.5-VL-72B
  41. 73.9 o1
  42. 73.7 Llama 4 Maverick
  43. 73.5 Command A Vision
  44. 73.1 Gemini 2.0 Flash
  45. 73.1 GPT-4.1 mini
  46. 73.1 BAGEL-7B-MoT
  47. 72.2 GPT-4.1
  48. 71.4 QVQ-72B-Preview
  49. 71.3 Kimi-VL-A3B-Thinking
  50. 70.7 Claude 3.5 Sonnet (October 2024)
  51. 70.7 Llama 4 Scout
  52. 70.5 Qwen2-VL-72B-Instruct
  53. 70.1 Magistral Medium 1.0
  54. 69.4 Pixtral Large
  55. 69.0 Grok-2
  56. 68.9 Mistral Small 3.1
  57. 68.5 LFM2.5-VL-3B
  58. 68.1 Grok-2 mini
  59. 67.9 Qwen2.5-Omni-7B
  60. 67.7 Claude 3.5 Sonnet
  61. 67.6 Gemma 3 27B
  62. 67.3 InternVL2-8B-MPO
  63. 67.2 Ovis1.6-Gemma2-9B
  64. 67.1 Mistral Small 3.2
  65. 66.8 InternVL2-Pro
  66. 66.8 Magistral Small 1.0
  67. 63.9 Gemini 1.5 Pro
  68. 63.8 GPT-4o
  69. 62.9 Gemma 3 12B
  70. 62.4 Phi-4-multimodal
  71. 62.2 Qwen3.5-0.8B
  72. 59.9 InternVL-Chat-V1-2-Plus
  73. 58.4 Gemini 1.5 Flash
  74. 58.1 GPT-4 Turbo
  75. 58.0 Pixtral 12B
  76. 57.6 InternLM-XComposer2-VL-7B
  77. 56.7 GPT-4o mini
  78. 56.2 GPT-4.1 nano
  79. 53.0 Gemini 1.0 Ultra
  80. 52.1 Gemini 1.5 Pro
  81. 51.7 LFM2-VL-1.6B
  82. 50.5 Claude 3 Opus
  83. 50.0 Gemma 3 4B
  84. 49.9 GPT-4V
  85. 47.9 Claude 3 Sonnet
  86. 47.7 InternVL-Chat-V1-2
  87. 46.5 LLaVA-v1.6-34B
  88. 46.4 Claude 3 Haiku
  89. 45.2 Gemini 1.0 Pro
  90. 44.5 Phi-3-vision-128k-instruct
  91. 43.9 Phi-3.5-vision-instruct
  92. 43.3 Qwen-VL-Plus
  93. 40.6 MiniCPM-V 2.0
  94. 35.9 MM1-7B-Chat
  95. 34.9 OmniLMM-12B
  96. 34.6 llava-v1.6-vicuna-7b
  97. 32.0 MM1-3B-Chat
  98. 27.6 LLaVA-v1.5-13B
  99. 25.3 instructblip-vicuna-7b
  100. 19.8 IDEFICS-9B-Instruct
  1. 1 Seed2.1 Pro 90.7%
  2. 2 Qwen3.5 397B-A17B 90.3%
  3. 3 Kimi K2.5 90.1%
  4. 4 Seed2.0 Pro 89.8%
  5. 5 MiniCPM-V-4.7-35B-A3B 88%
  6. 6 Qwen3.5-27B 87.8%
  7. 7 Seed1.8 87.7%
  8. 8 Qwen3.5-122B-A10B 87.4%
  9. 9 Qwen3.6 27B 87.4%
  10. 10 o3 86.8%
  11. 11 Qwen 3.6 35B-A3B 86.4%
  12. 12 Qwen3.5-35B-A3B 86.2%
  13. 13 Qwen3.5-Omni-Plus 86.1%
  14. 14 Qwen3-VL-235B-A22B 85.8%
  15. 15 Qwen3.5-9B 85.7%
  16. 16 Seed1.5-VL 85.6%
  17. 17 GLM-4.6V 85.2%
  18. 18 Qwen3.5-4B 85.1%
  19. 19 EXAONE 4.5 33B 85%
  20. 20 ERNIE 5.0 84.8%
  21. 21 GLM-4.5V 84.6%
  22. 22 o4-mini 84.3%
  23. 23 STEP3-VL-10B 84%
  24. 24 Keye-VL-2.0-30B-A3B 82.9%
  25. 25 Nemotron 3 Nano Omni 30B-A3B 82.8%
  26. 26 InternVL3.5-241B-A28B 82.7%
  27. 27 ERNIE-4.5-VL-28B-A3B-Thinking 82.5%
  28. 28 Intern-S1 81.5%
  29. 29 GLM-4.1V-9B-Thinking 80.7%
  30. 30 Command A+ 80.6%
  31. 31 Kimi-VL-A3B-Thinking-2506 80.1%
  32. 32 Doubao-1.5-pro 79.5%
  33. 33 LongCat-Flash-Omni 77.9%
  34. 34 Ovis2-34B 77.1%
  35. 35 Nemotron Nano 12B v2 VL 76.9%
  36. 36 Qwen3.5-2B 76.7%
  37. 37 Qwen3-Omni-30B-A3B 75.9%
  38. 38 Phi-4-Reasoning-Vision-15B 75.2%
  39. 39 Kimi k1.5 74.9%
  40. 40 Qwen2.5-VL-72B 74.8%
  41. 41 o1 73.9%
  42. 42 Llama 4 Maverick 73.7%
  43. 43 Command A Vision 73.5%
  44. 44 Gemini 2.0 Flash Feb 2025 73.1%
  45. 45 GPT-4.1 mini 73.1%
  46. 46 BAGEL-7B-MoT 73.1%
  47. 47 GPT-4.1 72.2%
  48. 48 QVQ-72B-Preview 71.4%
  49. 49 Kimi-VL-A3B-Thinking 71.3%
  50. 50 Claude 3.5 Sonnet (October 2024) 70.7%
  51. 51 Llama 4 Scout 70.7%
  52. 52 Qwen2-VL-72B-Instruct 70.5%
  53. 53 Magistral Medium 1.0 70.1%
  54. 54 Pixtral Large 69.4%
  55. 55 Grok-2 Aug 2024 69%
  56. 56 Mistral Small 3.1 68.9%
  57. 57 LFM2.5-VL-3B 68.5%
  58. 58 Grok-2 mini 68.1%
  59. 59 Qwen2.5-Omni-7B 67.9%
  60. 60 Claude 3.5 Sonnet 67.7%
  61. 61 Gemma 3 27B 67.6%
  62. 62 InternVL2-8B-MPO 67.3%
  63. 63 Ovis1.6-Gemma2-9B 67.2%
  64. 64 Mistral Small 3.2 67.1%
  65. 65 InternVL2-Pro 66.8%
  66. 66 Magistral Small 1.0 66.8%
  67. 67 Gemini 1.5 Pro May 2024 63.9%
  68. 68 GPT-4o May 2024 63.8%
  69. 69 Gemma 3 12B 62.9%
  70. 70 Phi-4-multimodal 62.4%
  71. 71 Qwen3.5-0.8B 62.2%
  72. 72 InternVL-Chat-V1-2-Plus 59.9%
  73. 73 Gemini 1.5 Flash May 2024 58.4%
  74. 74 GPT-4 Turbo Apr 2024 58.1%
  75. 75 Pixtral 12B 58%
  76. 76 InternLM-XComposer2-VL-7B 57.6%
  77. 77 GPT-4o mini 56.7%
  78. 78 GPT-4.1 nano 56.2%
  79. 79 Gemini 1.0 Ultra 53%
  80. 80 Gemini 1.5 Pro Feb 2024 52.1%
  81. 81 LFM2-VL-1.6B 51.7%
  82. 82 Claude 3 Opus 50.5%
  83. 83 Gemma 3 4B 50%
  84. 84 GPT-4V 49.9%
  85. 85 Claude 3 Sonnet 47.9%
  86. 86 InternVL-Chat-V1-2 47.7%
  87. 87 LLaVA-v1.6-34B 46.5%
  88. 88 Claude 3 Haiku 46.4%
  89. 89 Gemini 1.0 Pro 45.2%
  90. 90 Phi-3-vision-128k-instruct 44.5%
  91. 91 Phi-3.5-vision-instruct 43.9%
  92. 92 Qwen-VL-Plus 43.3%
  93. 93 MiniCPM-V 2.0 40.6%
  94. 94 MM1-7B-Chat 35.9%
  95. 95 OmniLMM-12B 34.9%
  96. 96 llava-v1.6-vicuna-7b 34.6%
  97. 97 MM1-3B-Chat 32%
  98. 98 LLaVA-v1.5-13B 27.6%
  99. 99 instructblip-vicuna-7b 25.3%
  100. 100 IDEFICS-9B-Instruct 19.8%

Sources: Labs’ model cards and official MathVista leaderboard

#ModelScoreSettingDateSource
1 Seed2.1 Pro 90.7% Jun 2026 Lab model card
2 Qwen3.5 397B-A17B 90.3% Feb 2026 Lab model card
3 Kimi K2.5 90.1% Jan 2026 Lab model card
4 Seed2.0 Pro 89.8% Feb 2026 Lab model card
5 MiniCPM-V-4.7-35B-A3B 88% Oct 2026 Lab model card
6 Qwen3.5-27B 87.8% Feb 2026 Lab model card
7 Seed1.8 87.7% Dec 2025 Lab model card
8 Qwen3.5-122B-A10B 87.4% Feb 2026 Lab model card
9 Qwen3.6 27B 87.4% Apr 2026 Lab model card
10 o3 86.8% Apr 2025 Lab model card
11 Qwen 3.6 35B-A3B 86.4% Apr 2026 Lab model card
12 Qwen3.5-35B-A3B 86.2% Feb 2026 Lab model card
13 Qwen3.5-Omni-Plus 86.1% Mar 2026 Lab model card
14 Qwen3-VL-235B-A22B 85.8% Sep 2025 Lab model card
15 Qwen3.5-9B 85.7% Mar 2026 Lab model card
16 Seed1.5-VL 85.6% May 2025 Lab model card
17 GLM-4.6V 85.2% Dec 2025 Lab model card
18 Qwen3.5-4B 85.1% Mar 2026 Lab model card
19 EXAONE 4.5 33B 85% Apr 2026 Lab model card
20 ERNIE 5.0 84.8% Feb 2026 Lab model card
21 GLM-4.5V 84.6% Aug 2025 Lab model card
22 o4-mini 84.3% Apr 2025 Lab model card
23 STEP3-VL-10B 84% Jan 2026 Lab model card
24 Keye-VL-2.0-30B-A3B 82.9% May 2026 Lab model card
25 Nemotron 3 Nano Omni 30B-A3B 82.8% Apr 2026 Lab model card
26 InternVL3.5-241B-A28B 82.7% Aug 2025 Lab model card
27 ERNIE-4.5-VL-28B-A3B-Thinking 82.5% Nov 2025 Lab model card
28 Intern-S1 81.5% Jul 2025 Lab model card
29 GLM-4.1V-9B-Thinking 80.7% Jul 2025 Lab model card
30 Command A+ 80.6% May 2026 Lab model card
31 Kimi-VL-A3B-Thinking-2506 80.1% Jun 2025 Lab model card
32 LongCat-Flash-Omni 77.9% Oct 2025 Lab model card
33 Nemotron Nano 12B v2 VL 76.9% Oct 2025 Lab model card
34 Qwen3.5-2B 76.7% Mar 2026 Lab model card
35 Qwen3-Omni-30B-A3B 75.9% Sep 2025 Lab model card
36 Phi-4-Reasoning-Vision-15B 75.2% Mar 2026 Lab model card
37 Kimi k1.5 74.9% Jan 2025 Lab model card
38 Qwen2.5-VL-72B 74.8% Jan 2025 Lab model card
39 Llama 4 Maverick 73.7% Apr 2025 Lab model card
40 Command A Vision 73.5% Jul 2025 Lab model card
41 GPT-4.1 mini 73.1% Apr 2025 Lab model card
42 BAGEL-7B-MoT 73.1% May 2025 Lab model card
43 GPT-4.1 72.2% Apr 2025 Lab model card
44 Kimi-VL-A3B-Thinking 71.3% Apr 2025 Lab model card
45 o1 71% Dec 2024 Lab model card
46 Claude 3.5 Sonnet (October 2024) 70.7% Oct 2024 Lab model card
47 Llama 4 Scout 70.7% Apr 2025 Lab model card
48 Magistral Medium 1.0 70.1% Jun 2025 Lab model card
49 Grok-2 Aug 2024 69% Aug 2024 Lab model card
50 Mistral Small 3.1 68.9% Mar 2025 Lab model card
51 LFM2.5-VL-3B 68.5% Aug 2026 Lab model card
52 Qwen2.5-Omni-7B 67.9% Mar 2025 Lab model card
53 Claude 3.5 Sonnet 67.7% Jun 2024 Lab model card
54 Gemma 3 27B 67.6% Mar 2025 Lab model card
55 Mistral Small 3.2 67.1% Jun 2025 Lab model card
56 Magistral Small 1.0 66.8% Jun 2025 Lab model card
57 Gemini 1.5 Pro May 2024 63.9% May 2024 Lab model card
58 GPT-4o May 2024 63.8% May 2024 Lab model card
59 Gemma 3 12B 62.9% Mar 2025 Lab model card
60 Phi-4-multimodal 62.4% Feb 2025 Lab model card
61 Qwen3.5-0.8B 62.2% Mar 2026 Lab model card
62 GPT-4o mini 56.7% Jul 2024 Lab model card
63 GPT-4.1 nano 56.2% Apr 2025 Lab model card
64 Gemini 1.0 Ultra 53% Dec 2023 Lab model card
65 Gemini 1.5 Pro Feb 2024 52.1% Feb 2024 Lab model card
66 LFM2-VL-1.6B 51.7% Nov 2025 Lab model card
67 Claude 3 Opus 50.5% Mar 2024 Lab model card
68 Gemma 3 4B 50% Mar 2025 Lab model card
69 Claude 3 Sonnet 47.9% Mar 2024 Lab model card
70 Claude 3 Haiku 46.4% Mar 2024 Lab model card
71 Phi-3.5-vision-instruct 43.9% Aug 2024 Lab model card
#ModelScoreSettingDateSource
1 Doubao-1.5-pro 79.5% Jan 2025 Official leaderboard
2 Ovis2-34B 77.1% Feb 2025 Official leaderboard
3 Kimi k1.5 74.9% Jan 2025 Official leaderboard
4 o1 73.9% Dec 2024 Official leaderboard
5 Llama 4 Maverick 73.7% Apr 2025 Official leaderboard
6 Gemini 2.0 Flash Feb 2025 73.1% Feb 2025 Official leaderboard
7 QVQ-72B-Preview 71.4% Dec 2024 Official leaderboard
8 Qwen2-VL-72B-Instruct 70.5% Aug 2024 Official leaderboard
9 Pixtral Large 69.4% Nov 2024 Official leaderboard
10 Grok-2 Aug 2024 69% Aug 2024 Official leaderboard
11 Grok-2 mini 68.1% Aug 2024 Official leaderboard
12 Claude 3.5 Sonnet 67.7% Jun 2024 Official leaderboard
13 InternVL2-8B-MPO 67.3% Nov 2024 Official leaderboard
14 Ovis1.6-Gemma2-9B 67.2% Sep 2024 Official leaderboard
15 InternVL2-Pro 66.8% Jul 2024 Official leaderboard
16 Gemini 1.5 Pro May 2024 63.9% May 2024 Official leaderboard
17 GPT-4o May 2024 63.8% May 2024 Official leaderboard
18 Phi-4-multimodal 62.4% Feb 2025 Official leaderboard
19 InternVL-Chat-V1-2-Plus 59.9% Feb 2024 Official leaderboard
20 Gemini 1.5 Flash May 2024 58.4% May 2024 Official leaderboard
21 GPT-4 Turbo Apr 2024 58.1% Apr 2024 Official leaderboard
22 Pixtral 12B 58% Sep 2024 Official leaderboard
23 InternLM-XComposer2-VL-7B 57.6% Jan 2024 Official leaderboard
24 Gemini 1.0 Ultra 53% Dec 2023 Official leaderboard
25 Gemini 1.5 Pro Feb 2024 52.1% Feb 2024 Official leaderboard
26 Claude 3 Opus 50.5% Feb 2024 Official leaderboard
27 GPT-4V 49.9% Nov 2023 Official leaderboard
28 Claude 3 Sonnet 47.9% Feb 2024 Official leaderboard
29 InternVL-Chat-V1-2 47.7% Feb 2024 Official leaderboard
30 LLaVA-v1.6-34B 46.5% Jan 2024 Official leaderboard
31 Claude 3 Haiku 46.4% Mar 2024 Official leaderboard
32 Gemini 1.0 Pro 45.2% Dec 2023 Official leaderboard
33 Phi-3-vision-128k-instruct 44.5% May 2024 Official leaderboard
34 Phi-3.5-vision-instruct 43.9% Aug 2024 Official leaderboard
35 Qwen-VL-Plus 43.3% Jan 2024 Official leaderboard
36 MiniCPM-V 2.0 40.6% Apr 2024 Official leaderboard
37 MM1-7B-Chat 35.9% Mar 2024 Official leaderboard
38 OmniLMM-12B 34.9% Jan 2024 Official leaderboard
39 llava-v1.6-vicuna-7b 34.6% Jan 2024 Official leaderboard
40 MM1-3B-Chat 32% Mar 2024 Official leaderboard
41 LLaVA-v1.5-13B 27.6% Oct 2023 Official leaderboard
42 instructblip-vicuna-7b 25.3% May 2023 Official leaderboard
43 IDEFICS-9B-Instruct 19.8% Jul 2023 Official leaderboard
#ModelScoreSettingDateSource
1 Seed2.1 Pro 90.7% Jun 2026 Lab model card
2 Qwen3.5 397B-A17B 90.3% Feb 2026 Lab model card
3 Kimi K2.5 90.1% Jan 2026 Lab model card
4 Seed2.0 Pro 89.8% Feb 2026 Lab model card
5 MiniCPM-V-4.7-35B-A3B 88% Oct 2026 Lab model card
6 Qwen3.5-27B 87.8% Feb 2026 Lab model card
7 Seed1.8 87.7% Dec 2025 Lab model card
8 Qwen3.5-122B-A10B 87.4% Feb 2026 Lab model card
9 Qwen3.6 27B 87.4% Apr 2026 Lab model card
10 o3 86.8% Apr 2025 Lab model card
11 Qwen 3.6 35B-A3B 86.4% Apr 2026 Lab model card
12 Qwen3.5-35B-A3B 86.2% Feb 2026 Lab model card
13 Qwen3.5-Omni-Plus 86.1% Mar 2026 Lab model card
14 Qwen3-VL-235B-A22B 85.8% Sep 2025 Lab model card
15 Qwen3.5-9B 85.7% Mar 2026 Lab model card
16 Seed1.5-VL 85.6% May 2025 Lab model card
17 GLM-4.6V 85.2% Dec 2025 Lab model card
18 Qwen3.5-4B 85.1% Mar 2026 Lab model card
19 EXAONE 4.5 33B 85% Apr 2026 Lab model card
20 ERNIE 5.0 84.8% Feb 2026 Lab model card
21 GLM-4.5V 84.6% Aug 2025 Lab model card
22 o4-mini 84.3% Apr 2025 Lab model card
23 STEP3-VL-10B 84% Jan 2026 Lab model card
24 Keye-VL-2.0-30B-A3B 82.9% May 2026 Lab model card
25 Nemotron 3 Nano Omni 30B-A3B 82.8% Apr 2026 Lab model card
26 InternVL3.5-241B-A28B 82.7% Aug 2025 Lab model card
27 ERNIE-4.5-VL-28B-A3B-Thinking 82.5% Nov 2025 Lab model card
28 Intern-S1 81.5% Jul 2025 Lab model card
29 GLM-4.1V-9B-Thinking 80.7% Jul 2025 Lab model card
30 Command A+ 80.6% May 2026 Lab model card
31 Kimi-VL-A3B-Thinking-2506 80.1% Jun 2025 Lab model card
32 Doubao-1.5-pro 79.5% Jan 2025 Official leaderboard
33 LongCat-Flash-Omni 77.9% Oct 2025 Lab model card
34 Ovis2-34B 77.1% Feb 2025 Official leaderboard
35 Nemotron Nano 12B v2 VL 76.9% Oct 2025 Lab model card
36 Qwen3.5-2B 76.7% Mar 2026 Lab model card
37 Qwen3-Omni-30B-A3B 75.9% Sep 2025 Lab model card
38 Phi-4-Reasoning-Vision-15B 75.2% Mar 2026 Lab model card
39 Kimi k1.5 74.9% Jan 2025 Official leaderboard
40 Qwen2.5-VL-72B 74.8% Jan 2025 Lab model card
41 o1 73.9% Dec 2024 Official leaderboard
42 Llama 4 Maverick 73.7% Apr 2025 Official leaderboard
43 Command A Vision 73.5% Jul 2025 Lab model card
44 Gemini 2.0 Flash Feb 2025 73.1% Feb 2025 Official leaderboard
45 GPT-4.1 mini 73.1% Apr 2025 Lab model card
46 BAGEL-7B-MoT 73.1% May 2025 Lab model card
47 GPT-4.1 72.2% Apr 2025 Lab model card
48 QVQ-72B-Preview 71.4% Dec 2024 Official leaderboard
49 Kimi-VL-A3B-Thinking 71.3% Apr 2025 Lab model card
50 Claude 3.5 Sonnet (October 2024) 70.7% Oct 2024 Lab model card
51 Llama 4 Scout 70.7% Apr 2025 Lab model card
52 Qwen2-VL-72B-Instruct 70.5% Aug 2024 Official leaderboard
53 Magistral Medium 1.0 70.1% Jun 2025 Lab model card
54 Pixtral Large 69.4% Nov 2024 Official leaderboard
55 Grok-2 Aug 2024 69% Aug 2024 Official leaderboard
56 Mistral Small 3.1 68.9% Mar 2025 Lab model card
57 LFM2.5-VL-3B 68.5% Aug 2026 Lab model card
58 Grok-2 mini 68.1% Aug 2024 Official leaderboard
59 Qwen2.5-Omni-7B 67.9% Mar 2025 Lab model card
60 Claude 3.5 Sonnet 67.7% Jun 2024 Official leaderboard
61 Gemma 3 27B 67.6% Mar 2025 Lab model card
62 InternVL2-8B-MPO 67.3% Nov 2024 Official leaderboard
63 Ovis1.6-Gemma2-9B 67.2% Sep 2024 Official leaderboard
64 Mistral Small 3.2 67.1% Jun 2025 Lab model card
65 InternVL2-Pro 66.8% Jul 2024 Official leaderboard
66 Magistral Small 1.0 66.8% Jun 2025 Lab model card
67 Gemini 1.5 Pro May 2024 63.9% May 2024 Official leaderboard
68 GPT-4o May 2024 63.8% May 2024 Official leaderboard
69 Gemma 3 12B 62.9% Mar 2025 Lab model card
70 Phi-4-multimodal 62.4% Feb 2025 Official leaderboard
71 Qwen3.5-0.8B 62.2% Mar 2026 Lab model card
72 InternVL-Chat-V1-2-Plus 59.9% Feb 2024 Official leaderboard
73 Gemini 1.5 Flash May 2024 58.4% May 2024 Official leaderboard
74 GPT-4 Turbo Apr 2024 58.1% Apr 2024 Official leaderboard
75 Pixtral 12B 58% Sep 2024 Official leaderboard
76 InternLM-XComposer2-VL-7B 57.6% Jan 2024 Official leaderboard
77 GPT-4o mini 56.7% Jul 2024 Lab model card
78 GPT-4.1 nano 56.2% Apr 2025 Lab model card
79 Gemini 1.0 Ultra 53% Dec 2023 Official leaderboard
80 Gemini 1.5 Pro Feb 2024 52.1% Feb 2024 Official leaderboard
81 LFM2-VL-1.6B 51.7% Nov 2025 Lab model card
82 Claude 3 Opus 50.5% Feb 2024 Official leaderboard
83 Gemma 3 4B 50% Mar 2025 Lab model card
84 GPT-4V 49.9% Nov 2023 Official leaderboard
85 Claude 3 Sonnet 47.9% Feb 2024 Official leaderboard
86 InternVL-Chat-V1-2 47.7% Feb 2024 Official leaderboard
87 LLaVA-v1.6-34B 46.5% Jan 2024 Official leaderboard
88 Claude 3 Haiku 46.4% Mar 2024 Official leaderboard
89 Gemini 1.0 Pro 45.2% Dec 2023 Official leaderboard
90 Phi-3-vision-128k-instruct 44.5% May 2024 Official leaderboard
91 Phi-3.5-vision-instruct 43.9% Aug 2024 Official leaderboard
92 Qwen-VL-Plus 43.3% Jan 2024 Official leaderboard
93 MiniCPM-V 2.0 40.6% Apr 2024 Official leaderboard
94 MM1-7B-Chat 35.9% Mar 2024 Official leaderboard
95 OmniLMM-12B 34.9% Jan 2024 Official leaderboard
96 llava-v1.6-vicuna-7b 34.6% Jan 2024 Official leaderboard
97 MM1-3B-Chat 32% Mar 2024 Official leaderboard
98 LLaVA-v1.5-13B 27.6% Oct 2023 Official leaderboard
99 instructblip-vicuna-7b 25.3% May 2023 Official leaderboard
100 IDEFICS-9B-Instruct 19.8% Jul 2023 Official leaderboard

The benchmark's own leaderboard has 76 entries, 43 of them models we track. Switch the scores above to rank them together.

Rankings by part

We read 12 parts from the benchmark’s own leaderboard in 2 groups: task type and domain. Each ranks models on one part of the test. Doubao-1.5-pro leads 10 of them, and Ovis2-34B leads the other 2. Arrows show how far a model moved from its overall rank among the models we track.

Figure question answering

15 models we track have a score here.

  1. 1 Doubao-1.5-pro 77.7%
  2. 2 Ovis2-34B 75.5%
  3. 3 InternVL2-8B-MPO ↑10 72.5%
  4. 4 InternVL2-Pro ↑11 70.6%
  5. 5 InternLM-XComposer2-VL-7B ↑18 55%
  6. 6 Qwen-VL-Plus ↑29 54.6%
  7. 7 MiniCPM-V 2.0 ↑29 53.2%
  8. 8 InternVL-Chat-V1-2-Plus ↑11 51.7%
  9. 9 InternVL-Chat-V1-2 ↑20 50.9%
  10. 10 Gemini 1.0 Ultra ↑14 49.1%

13 models have held number one in labs’ own numbers since its release in October 2023 14 models have held number one across all independent runs since its release in October 2023 20 models have held number one with claims included since its release in October 2023

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.

Other labs University of Wisconsin-Madison OpenAI Google DeepMind Shanghai AI Lab Anthropic xAI Moonshot AI ByteDance Seed Qwen
Own leaderboard Not tested since Apr 2025
Lab claims Passed 85% in Apr 2025, by o3
Seed2.1 Pro 90.7%

Sources: Official MathVista leaderboard and labs’ model cards

OpenAI’s own number for o3 passed 85% in April 2025

That was 18 months after its release. By the same age, 29% of the benchmarks from 2023 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%) Seed2.1 Pro o3 Doubao-1.5-pro Kimi k1.5 Claude 3.5 Sonnet (October 2024) Gemini 1.5 Pro InternLM-XComposer2-VL-7B 0% 20% 40% 60% 80% 100% 2023 2024 2025 2026 Near the ceiling (85%) Seed2.1 Pro Doubao-1.5-pro

Sources: Official MathVista leaderboard and labs’ model cards

2 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.

Google DeepMind
Anthropic
OpenAI
xAI
Moonshot AI
Qwen
Microsoft
Baidu
Mistral AI
Meta
ByteDance Seed
Z.ai
Shanghai AI Lab
Cohere
Liquid AI
NVIDIA
Meituan
StepFun
LG AI Research
Kuaishou
OpenBMB (ModelBest)
DateLaunch postScoreAs the post writes it
Oct 7, 2026 MiniCPM-V-4.7-35B-A3B 88 MathVista
Aug 12, 2026 LFM2.5-VL-3B 68.5 MathVista (mini)
May 25, 2026 Keye-VL-2.0-30B-A3B 82.9 Mathvista (mini)
May 20, 2026 Command A+ 80.6 MathVista
Apr 28, 2026 Nemotron 3 Nano Omni 30B-A3B 82.8 MathVista_MINI
Apr 22, 2026 Qwen3.6 27B 87.4 MathVista mini
Apr 15, 2026 Qwen 3.6 35B-A3B 86.4 Mathvista(mini)
Apr 9, 2026 EXAONE 4.5 33B 85 MathVista (mini)
Mar 30, 2026 Qwen3.5-Omni-Plus 86.1 Mathvista (mini)
Mar 4, 2026 Phi-4-Reasoning-Vision-15B 75.2 MathVista_MINI (default mode)
Mar 2, 2026 Qwen3.5-0.8B 62.2 Mathvista(mini) 62.2/58.6 (with CI / without CI)
Mar 2, 2026 Qwen3.5-2B 76.7 Mathvista(mini) 76.7/73.9 (with CI / without CI)
Mar 2, 2026 Qwen3.5-4B 85.1 Mathvista(mini)
Mar 2, 2026 Qwen3.5-9B 85.7 Mathvista(mini)
Feb 24, 2026 Qwen3.5-35B-A3B 86.2 Mathvista(mini)
Feb 24, 2026 Qwen3.5-27B 87.8 Mathvista(mini)
Feb 24, 2026 Qwen3.5-122B-A10B 87.4 Mathvista(mini)
Feb 16, 2026 Qwen3.5 397B-A17B 90.3 Mathvista(mini)
Feb 14, 2026 Seed2.0 Pro 89.8 MathVista
Feb 6, 2026 ERNIE 5.0 84.8 MathVista
Jan 27, 2026 Kimi K2.5 90.1 MathVista (mini)
Jan 13, 2026 STEP3-VL-10B 84 MathVista (SeRe)
Dec 18, 2025 Seed1.8 87.7 MathVista
Dec 8, 2025 GLM-4.6V 85.2 MathVista
Nov 11, 2025 ERNIE-4.5-VL-28B-A3B-Thinking 82.5 MathVista
Oct 31, 2025 LongCat-Flash-Omni 77.9 MathVista mini
Oct 28, 2025 Nemotron Nano 12B v2 VL 76.9 MathVista*
Sep 23, 2025 Qwen3-VL-235B-A22B 85.8 MathVista_mini
Sep 22, 2025 Qwen3-Omni-30B-A3B 75.9 MathVista_mini
Aug 25, 2025 InternVL3.5-241B-A28B 82.7 MathVista
Aug 12, 2025 LFM2-VL-1.6B 51.1 MathVista
Aug 11, 2025 GLM-4.5V 84.6 MathVista
Jul 31, 2025 Command A Vision 73.5 MathVista
Jul 24, 2025 Intern-S1 81.5 MathVista
Jul 1, 2025 GLM-4.1V-9B-Thinking 80.7 MathVista
Jun 21, 2025 Kimi-VL-A3B-Thinking-2506 80.1 MathVista_MINI (Pass@1)
Jun 20, 2025 Mistral Small 3.2 67.1 Mathvista
May 20, 2025 BAGEL-7B-MoT 73.1 MathVista
May 12, 2025 Seed1.5-VL 85.6 MathVista
Apr 16, 2025 o4-mini 84.3 MathVista
Apr 16, 2025 o3 86.8 MathVista Visual Math Reasoning
Apr 14, 2025 GPT-4.1 nano 56.2 MathVista
Apr 14, 2025 GPT-4.1 mini 73.1 MathVista
Apr 14, 2025 GPT-4.1 72.2 MathVista
Apr 10, 2025 Kimi-VL-A3B-Thinking 71.3 MathVista (mini) (Pass@1)
Apr 5, 2025 Llama 4 Scout 70.7 MathVista
Apr 5, 2025 Llama 4 Maverick 73.7 MathVista
Mar 27, 2025 Qwen2.5-Omni-7B 67.9 MathVistatestmini
Mar 17, 2025 Mistral Small 3.1 68.9 Mathvista
Mar 16, 2025 ERNIE 4.5 MathVista
Mar 12, 2025 Gemma 3 27B 67.6 MathVista (testmini) (IT)
Feb 26, 2025 Phi-4-multimodal 62.4 MathVista (testmini)
Jan 28, 2025 Qwen2.5-VL-72B 74.8 MathVista_MINI
Jan 20, 2025 Kimi k1.5 74.9 MathVista-Test (Pass@1), long-CoT
Dec 17, 2024 o1 71 MathVista (pass @1)
Oct 22, 2024 Claude 3.5 Sonnet (October 2024) 70.7 MathVista (testmini)
Sep 12, 2024 o1-preview MathVista (testmini) pass@1
Aug 13, 2024 Grok-2 69 MathVista (0-shot CoT)
Jul 18, 2024 GPT-4o mini 56.7 MathVista
Jun 20, 2024 Claude 3.5 Sonnet 67.7 MathVista (testmini)
May 14, 2024 Gemini 1.5 Pro 63.9 MathViSTA (testmini), 0-shot
May 13, 2024 GPT-4o 63.8 MathVista (%) (testmini)
Mar 4, 2024 Claude 3 Opus 50.5 MathVista (testmini) CoT
Feb 15, 2024 Gemini 1.5 Pro 52.1 MathViSTA (testmini), 0-shot
Dec 6, 2023 Gemini 1.0 Ultra 53 MathVista (testmini)

42 system cards report it, from 17 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 23, 2026 Seed2.1 Pro model card 90.7 MathVista (Table 9) Capabilities
Jun 9, 2026 Keye-VL-2.0-30B-A3B technical report 82.9 MathVista (mini) (Figure 5, printed value) Capabilities
Apr 27, 2026 Nemotron 3 Nano Omni 30B-A3B technical report 82.8 MathVista-Mini, reasoning on, 71.9 reasoning off (Table 7) Capabilities
Apr 17, 2026 Qwen3.5-Omni-Plus technical report 86.1 Mathvista (mini) (Table 6) Capabilities
Apr 9, 2026 EXAONE 4.5 33B technical report 85 MathVista (mini) (Table 2, Reasoning mode) Capabilities
Mar 21, 2026 Seed1.8 model card 87.7 MathVista (Table 2) Capabilities
Mar 4, 2026 Phi-4-Reasoning-Vision-15B technical report 75.2 MathVista MINI, default mode, 68.7 force nothink, 74.1 force thinking (Tables 5 and 6) Capabilities
Feb 14, 2026 Seed2.0 Pro model card 89.8 MathVista (testmini) (Table 8) Capabilities
Feb 4, 2026 ERNIE 5.0 technical report 84.8 MathVista (Table 3) Capabilities
Jan 30, 2026 Kimi K2.5 technical report 90.1 MathVista (mini) (Table 4) Capabilities
Jan 14, 2026 STEP3-VL-10B technical report 84 MathVista, SeRe (Tables 1 and 3). PaCoRe 85.50 Capabilities
Jan 1, 2026 GLM-4.6V technical report 85.2 MathVista, GLM-4.6V thinking (Table 2) Capabilities
Nov 28, 2025 LFM2-VL-1.6B technical report 51.7 MathVista, LFM2-VL-1.6B (Table 8) Capabilities
Nov 26, 2025 Qwen3-VL-235B-A22B technical report 85.8 MathVista mini (Table 2) Capabilities
Nov 6, 2025 Nemotron Nano 12B v2 VL technical report 75.5 MathVista-Mini, reasoning on (Table 3). Reasoning off 69.0 Capabilities
Oct 31, 2025 LongCat-Flash-Omni technical report 77.9 MathVista mini (Table 5) Capabilities
Sep 22, 2025 Qwen3-Omni-30B-A3B technical report 75.9 MathVista mini, Instruct (Table 9). Thinking 80.0 (Table 10) Capabilities
Aug 25, 2025 InternVL3.5-241B-A28B technical report 82.7 MathVista (mini) (Table 2, Table 3). With parallel thinking 84.8 Capabilities
Aug 21, 2025 Intern-S1 technical report 81.5 MathVista (Table 2) Capabilities
Aug 13, 2025 GLM-4.5V technical report 84.6 MathVista, GLM-4.5V thinking (Table 2) Capabilities
Jul 1, 2025 GLM-4.1V-9B-Thinking technical report 80.7 MathVista, GLM-4.1V-9B-Thinking thinking (Table 2) Capabilities
Jun 23, 2025 Kimi-VL-A3B-Thinking-2506 technical report 80.1 MathVista (mini) (Pass@1) (Table 4) Capabilities
Jun 12, 2025 Magistral Small 1.0 technical report 66.8 MathVista, Magistral Small bar, printed label (Figure 10) Capabilities
Jun 12, 2025 Magistral Medium 1.0 technical report 70.1 MathVista, printed bar label (Figure 10) Capabilities
May 20, 2025 BAGEL-7B-MoT technical report 73.1 MathVista (Table 4) Capabilities
May 11, 2025 Seed1.5-VL technical report 85.6 MathVista, thinking (Table 6). Non-thinking 83.0 Capabilities
Apr 10, 2025 Kimi-VL-A3B-Thinking technical report 71.3 MathVista (mini) (Pass@1) (Table 4). Figure 13 shows 70.9 at 4k max thinking tokens Capabilities
Apr 5, 2025 Llama 4 Scout model card 70.7 MathVista, 0-shot, accuracy, instruction tuned (Instruction tuned models table) Capabilities
Apr 5, 2025 Llama 4 Maverick model card 73.7 MathVista, 0-shot, accuracy, instruction tuned (Instruction tuned models table) Capabilities
Mar 26, 2025 Qwen2.5-Omni-7B technical report 67.9 MathVista testmini (Table 5) Capabilities
Mar 25, 2025 Gemma 3 4B technical report 50 MathVista (testmini), IT, P&S applied (Table 16) Capabilities
Mar 25, 2025 Gemma 3 27B technical report 67.6 MathVista (testmini), IT, P&S applied (Table 16) Capabilities
Mar 25, 2025 Gemma 3 12B technical report 62.9 MathVista (testmini), IT, P&S applied (Table 16) Capabilities
Mar 3, 2025 Phi-4-multimodal technical report 62.4 MathVista (testmini) (Table 1) Capabilities
Feb 19, 2025 Qwen2.5-VL-72B technical report 74.8 MathVista mini (Table 3) Capabilities
Jan 20, 2025 Kimi k1.5 technical report 74.9 MathVista-Test (Pass@1), long-CoT (Table 2). Short-CoT 70.1 (Table 3) Capabilities
Oct 22, 2024 Claude 3.5 Sonnet (October 2024) model card 70.7 MathVista (testmini), 0-shot CoT (Table 5) Capabilities
Aug 30, 2024 Phi-3.5-vision-instruct technical report 43.9 MathVista (testmini) (Table 5) Capabilities
Jun 20, 2024 Claude 3.5 Sonnet model card 67.7 MathVista (testmini), 0-shot CoT (Table 2) Capabilities
Mar 4, 2024 Claude 3 Sonnet model card 47.9 MathVista (testmini), 0-shot CoT (Table 3) Capabilities
Mar 4, 2024 Claude 3 Opus model card 50.5 MathVista (testmini), 0-shot CoT (Table 3) Capabilities
Mar 4, 2024 Claude 3 Haiku model card 46.4 MathVista (testmini), 0-shot CoT (Table 3) Capabilities

14 lab claims next to the leaderboard

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. 13 of them match, which usually means one side copied the other. Every benchmark.

Model Lab says Board says Gap
o1 71% MathVista (pass @1) 73.9% Leaderboard −2.9 pts
Gemini 1.0 Ultra 53% MathVista (testmini) 53% Leaderboard Same
Gemini 1.5 Pro Feb 2024 52.1% MathViSTA (testmini), 0-shot 52.1% Leaderboard Same
Claude 3 Opus 50.5% MathVista (testmini) CoT 50.5% Leaderboard Same
GPT-4o May 2024 63.8% MathVista (%) (testmini) 63.8% Leaderboard Same
Gemini 1.5 Pro May 2024 63.9% MathViSTA (testmini), 0-shot 63.9% Leaderboard Same
Claude 3.5 Sonnet 67.7% MathVista (testmini) 67.7% Leaderboard Same
Grok-2 Aug 2024 69% MathVista (0-shot CoT) 69% Leaderboard Same
Kimi k1.5 74.9% MathVista-Test (Pass@1), long-CoT 74.9% Leaderboard Same
Phi-4-multimodal 62.4% MathVista (testmini) 62.4% Leaderboard Same
Llama 4 Maverick 73.7% MathVista 73.7% Leaderboard Same
Claude 3 Haiku 46.4% MathVista (testmini), 0-shot CoT (Table 3) 46.4% Leaderboard Same
Claude 3 Sonnet 47.9% MathVista (testmini), 0-shot CoT (Table 3) 47.9% Leaderboard Same
Phi-3.5-vision-instruct 43.9% MathVista (testmini) (Table 5) 43.9% Leaderboard Same

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%
  • MATH-Vision (MathVision) In use 3,040 competition math problems with visual context. Answer accuracy. Qwen 3.8 Max 97.7%
  • MMStar In use Models answer 1,500 human-curated image questions that cannot be solved without the image, across 6 core capabilities. Scored by accuracy. Qwen 3.8 Max 85.9%
  • 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 (78 entries), launch posts (63 scores) and system cards (42 scores).
  • Pass Models scored 100 models we track, plus 35 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 Kimi K2.5 (#76).
  • Note Headroom Saturated. The best is 90.7%, past 85% since April 2025.
Leaderboard import 3 pass
  • Pass Importer We read the board on mathvista.github.io: 45 rows for models we track and 33 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 All 14 pairs agree within 10 points.
  • Pass Known issues No published audit found problems.

How we check