Benchmark Almanac

Vision Arena

Claude Fable 5 1,309 leads since June 2026
Not in model cards

People send an image with a prompt, compare answers from two anonymous vision-language models, and vote for the better one. Votes become Bradley-Terry ratings.

Launched as the Multimodal Arena on June 27th, 2024 with 17,429 votes. Since September 2025 an LLM filter drops image-generation prompts.

Released
June 2024
Kind
General visual questions
Level
Everyday
Scored
Ranked by votes
Answers
Open-ended
Strongest model tested
GPT-6 Astra #3 on the capability index
Last model added
October 2026
Scores refreshed
November 12th, 2026
153
Entries on its own leaderboard, 136 models we track

Claude Fable 5 leads with 1,309

Best score per entry from the benchmark's own leaderboard, read on November 12th, 2026. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders.

153 entries. Scroll sideways for the rest. Bars start at 800.

  1. 1,309 Claude Fable 5
  2. 1,300 Claude Opus 4.7
  3. 1,299 Qwen 3.8 Max
  4. 1,299 Claude Opus 4.6
  5. 1,295 Claude Opus 5.5
  6. 1,294 Muse Spark
  7. 1,294 Gemini 3.7 Flash
  8. 1,294 Muse Spark 1.2
  9. 1,293 Gemini 3.8 Flash
  10. 1,293 GPT-6.1 Sol
  11. 1,292 Muse Spark 1.3
  12. 1,289 Gemini 3 Pro
  13. 1,289 Claude Fable 5.1
  14. 1,288 GPT-5.6 Sol
  15. 1,286 Claude Opus 4.8
  16. 1,285 Claude Opus 5
  17. 1,285 GPT-5.5
  18. 1,283 GPT-6 Astra
  19. 1,282 GPT-5.4
  20. 1,281 Gemini 3.5 Flash
  21. 1,280 Grok 4.5
  22. 1,280 Gemini 3.1 Pro
  23. 1,280 Gemini 3.6 Flash
  24. 1,279 Muse Spark 1.1
  25. 1,278 GPT-5.2 Instant
  26. 1,277 GPT-5.5 Instant
  27. 1,275 Claude Sonnet 4.6
  28. 1,272 Gemini 3 Flash
  29. 1,272 GLM-5.3-Flash
  30. 1,266 GPT-5.6 Terra
  31. 1,266 Gemini 3.5 Flash-Lite
  32. 1,265 Gemma 4 31B IT
  33. 1,265 Kimi K2.6
  34. 1,263 Grok 4.6
  35. 1,263 DeepSeek V4.1 Flash
  36. 1,263 Claude Sonnet 5.5
  37. 1,262 Qwen3.7-Plus
  38. 1,261 Claude Sonnet 5
  39. 1,258 GPT-6 Sol
  40. 1,256 Seed2.0 Pro
  41. 1,256 Grok 4.20 Beta (Reasoning, 0309)
  42. 1,255 GPT-5.6 Luna
  43. 1,252 Kimi K2.5
  44. 1,251 GPT-5.1
  45. 1,251 Grok 4.20 Multi-Agent Beta (0309)
  46. 1,250 GPT-5.4 Mini
  47. 1,249 Grok 4.7
  48. 1,248 MiMo-V2.6 Flash
  49. 1,247 Gemini 2.5 Pro
  50. 1,247 MiMo-V2.6 Pro
  51. 1,246 Qwen3.5 397B-A17B
  52. 1,245 Step 5 Preview
  53. 1,243 Qwen 3.8 27B
  54. 1,242 GPT-5.2
  55. 1,242 Gemma 4 26B A4B
  56. 1,242 Grok 4.3
  57. 1,238 GPT-4o
  58. 1,235 GPT-6 Luna
  59. 1,235 Gemini 3.1 Flash-Lite
  60. 1,234 MiMo-V2.5
  61. 1,232 MiniMax-M3
  62. 1,232 GLM-5V-Turbo
  63. 1,231 Mistral Large 4
  64. 1,229 Gemini 2.5 Flash
  65. 1,226 Qwen3.5-122B-A10B
  66. 1,225 GPT-5 Chat
  67. 1,225 GPT-4.5
  68. 1,219 ERNIE-5.0-Preview (1220)
  69. 1,216 MiMo-V2-Omni
  70. 1,216 Qwen3.5-27B
  71. 1,216 o3
  72. 1,215 Gemini 2.5 Flash
  73. 1,214 Qwen3-VL-235B-A22B-Instruct
  74. 1,214 GPT-4.1
  75. 1,212 Nemotron 3.5 Super VL
  76. 1,211 GPT-5
  77. 1,208 Claude Opus 4
  78. 1,208 Inkling-Small
  79. 1,206 Claude Sonnet 4
  80. 1,201 o4-mini
  81. 1,201 GPT-4.1 mini
  82. 1,200 Mistral Large 3
  83. 1,199 Mistral Medium 3.5
  84. 1,199 GPT-5.4 Nano
  85. 1,196 Grok 4.1 Fast
  86. 1,193 o1
  87. 1,193 Claude 3.7 Sonnet
  88. 1,189 Gemini 2.5 Flash-Lite
  89. 1,188 Qwen3-VL-235B-A22B
  90. 1,185 GPT-5 mini
  91. 1,185 Qwen-VL-Max
  92. 1,184 Grok 4
  93. 1,180 Gemini 1.5 Pro
  94. 1,174 Gemini 2.5 Flash-Lite
  95. 1,172 Gemini 2.0 Flash
  96. 1,162 GPT-4o
  97. 1,161 Claude 3.5 Sonnet (October 2024)
  98. 1,161 GLM-4.6V
  99. 1,160 Gemma 3 27B
  100. 1,160 Mistral Medium 3.1
  101. 1,155 Step-1o Turbo (202506)
  102. 1,155 Mistral Medium 3
  103. 1,154 Hunyuan Vision 1.5 Thinking
  104. 1,153 GLM-4.5V
  105. 1,150 Hunyuan Large Vision
  106. 1,147 GPT-5 nano
  107. 1,146 Claude 3.5 Sonnet
  108. 1,145 Llama 4 Maverick
  109. 1,144 Step 3
  110. 1,142 Gemini 1.5 Flash
  111. 1,140 Mistral Small 3.2
  112. 1,136 Gemini 2.0 Flash-Lite
  113. 1,128 Llama 4 Scout
  114. 1,127 Claude 3.5 Haiku
  115. 1,127 Mistral Small 3.1
  116. 1,124 Step-1o Vision 32K (High Res)
  117. 1,121 Qwen2.5-VL-72B
  118. 1,120 Gemini 1.5 Pro
  119. 1,118 Qwen2.5-VL-32B
  120. 1,118 GPT-4o
  121. 1,113 Molmo2-8B
  122. 1,112 GPT-4 Turbo
  123. 1,097 GPT-4o mini
  124. 1,095 Pixtral Large (2411)
  125. 1,089 GPT-4.1 nano
  126. 1,085 Qwen2-VL-72B
  127. 1,085 Qwen-VL-Max (1119)
  128. 1,071 Gemini 1.5 Flash 8B
  129. 1,064 Step-1V 32K
  130. 1,062 Claude 3 Opus
  131. 1,059 Gemini 1.5 Flash
  132. 1,045 Molmo-72B
  133. 1,042 Hunyuan Standard Vision 20241231
  134. 1,032 Llama 3.2 90B
  135. 1,031 Qwen2-VL-7B-Instruct
  136. 1,025 Pixtral 12B (2409)
  137. 1,025 InternVL2-26B
  138. 1,019 Amazon Nova Lite
  139. 1,019 Amazon Nova Pro
  140. 1,017 Claude 3 Sonnet
  141. 1,003 Yi Vision
  142. 1,001 Aya Vision 32B
  143. 1,000 Claude 3 Haiku
  144. 994 Molmo-7B-D
  145. 993 Llama 3.2 11B
  146. 986 nvila-internal-15b-v1
  147. 978 LLaVA-OneVision Qwen2 72B OV
  148. 965 LLaVA 1.6 34B
  149. 964 CogVLM2-19B-Chat
  150. 963 MiniCPM-V-2_6
  151. 959 InternVL2-4B
  152. 919 Phi-3.5-vision-instruct
  153. 881 Phi-3-vision-128k-instruct

Bars start at 800.

  1. 1 Claude Fable 5 high 1,309
  2. 2 Claude Opus 4.7 1,300
  3. 3 Qwen 3.8 Max 1,299
  4. 4 Claude Opus 4.6 high 1,299
  5. 5 Claude Opus 5.5 high 1,295
  6. 6 Muse Spark 1,294
  7. 7 Gemini 3.7 Flash high 1,294
  8. 8 Muse Spark 1.2 xhigh 1,294
  9. 9 Gemini 3.8 Flash high 1,293
  10. 10 GPT-6.1 Sol max 1,293
  11. 11 Muse Spark 1.3 max 1,292
  12. 12 Gemini 3 Pro 1,289
  13. 13 Claude Fable 5.1 max 1,289
  14. 14 GPT-5.6 Sol xhigh 1,288
  15. 15 Claude Opus 4.8 high 1,286
  16. 16 Claude Opus 5 high 1,285
  17. 17 GPT-5.5 1,285
  18. 18 GPT-6 Astra max 1,283
  19. 19 GPT-5.4 high 1,282
  20. 20 Gemini 3.5 Flash medium 1,281
  21. 21 Grok 4.5 1,280
  22. 22 Gemini 3.1 Pro 1,280
  23. 23 Gemini 3.6 Flash high 1,280
  24. 24 Muse Spark 1.1 1,279
  25. 25 GPT-5.2 Instant 1,278
  26. 26 GPT-5.5 Instant 1,277
  27. 27 Claude Sonnet 4.6 1,275
  28. 28 Gemini 3 Flash 1,272
  29. 29 GLM-5.3-Flash 1,272
  30. 30 GPT-5.6 Terra xhigh 1,266
  31. 31 Gemini 3.5 Flash-Lite 1,266
  32. 32 Gemma 4 31B IT 1,265
  33. 33 Kimi K2.6 1,265
  34. 34 Grok 4.6 high 1,263
  35. 35 DeepSeek V4.1 Flash max 1,263
  36. 36 Claude Sonnet 5.5 xhigh 1,263
  37. 37 Qwen3.7-Plus 1,262
  38. 38 Claude Sonnet 5 high 1,261
  39. 39 GPT-6 Sol max 1,258
  40. 40 Seed2.0 Pro 1,256
  41. 41 Grok 4.20 Beta (Reasoning, 0309) 1,256
  42. 42 GPT-5.6 Luna xhigh 1,255
  43. 43 Kimi K2.5 thinking 1,252
  44. 44 GPT-5.1 high 1,251
  45. 45 Grok 4.20 Multi-Agent Beta (0309) 1,251
  46. 46 GPT-5.4 Mini high 1,250
  47. 47 Grok 4.7 xhigh 1,249
  48. 48 MiMo-V2.6 Flash 1,248
  49. 49 Gemini 2.5 Pro Jun 2025 1,247
  50. 50 MiMo-V2.6 Pro 1,247
  51. 51 Qwen3.5 397B-A17B 1,246
  52. 52 Step 5 Preview 1,245
  53. 53 Qwen 3.8 27B 1,243
  54. 54 GPT-5.2 high 1,242
  55. 55 Gemma 4 26B A4B 1,242
  56. 56 Grok 4.3 1,242
  57. 57 GPT-4o Mar 2025 1,238
  58. 58 GPT-6 Luna max 1,235
  59. 59 Gemini 3.1 Flash-Lite 1,235
  60. 60 MiMo-V2.5 1,234
  61. 61 MiniMax-M3 1,232
  62. 62 GLM-5V-Turbo 1,232
  63. 63 Mistral Large 4 1,231
  64. 64 Gemini 2.5 Flash Sep 2025 1,229
  65. 65 Qwen3.5-122B-A10B 1,226
  66. 66 GPT-5 Chat 1,225
  67. 67 GPT-4.5 1,225
  68. 68 ERNIE-5.0-Preview (1220) 1,219
  69. 69 MiMo-V2-Omni 1,216
  70. 70 Qwen3.5-27B 1,216
  71. 71 o3 1,216
  72. 72 Gemini 2.5 Flash Jun 2025 1,215
  73. 73 Qwen3-VL-235B-A22B-Instruct 1,214
  74. 74 GPT-4.1 1,214
  75. 75 Nemotron 3.5 Super VL nvfp4 1,212
  76. 76 GPT-5 high 1,211
  77. 77 Claude Opus 4 thinking 16k 1,208
  78. 78 Inkling-Small 1,208
  79. 79 Claude Sonnet 4 thinking 32k 1,206
  80. 80 o4-mini 1,201
  81. 81 GPT-4.1 mini 1,201
  82. 82 Mistral Large 3 1,200
  83. 83 Mistral Medium 3.5 1,199
  84. 84 GPT-5.4 Nano high 1,199
  85. 85 Grok 4.1 Fast reasoning 1,196
  86. 86 o1 1,193
  87. 87 Claude 3.7 Sonnet thinking 32k 1,193
  88. 88 Gemini 2.5 Flash-Lite Jun 2025 1,189
  89. 89 Qwen3-VL-235B-A22B thinking 1,188
  90. 90 GPT-5 mini high 1,185
  91. 91 Qwen-VL-Max 1,185
  92. 92 Grok 4 1,184
  93. 93 Gemini 1.5 Pro Sep 2024 1,180
  94. 94 Gemini 2.5 Flash-Lite Sep 2025, non-thinking, 09-2025 1,174
  95. 95 Gemini 2.0 Flash Feb 2025 1,172
  96. 96 GPT-4o May 2024 1,162
  97. 97 Claude 3.5 Sonnet (October 2024) 1,161
  98. 98 GLM-4.6V 1,161
  99. 99 Gemma 3 27B 1,160
  100. 100 Mistral Medium 3.1 1,160
  101. 101 Step-1o Turbo (202506) 1,155
  102. 102 Mistral Medium 3 1,155
  103. 103 Hunyuan Vision 1.5 Thinking 1,154
  104. 104 GLM-4.5V 1,153
  105. 105 Hunyuan Large Vision 1,150
  106. 106 GPT-5 nano high 1,147
  107. 107 Claude 3.5 Sonnet 1,146
  108. 108 Llama 4 Maverick 1,145
  109. 109 Step 3 1,144
  110. 110 Gemini 1.5 Flash Sep 2024 1,142
  111. 111 Mistral Small 3.2 1,140
  112. 112 Gemini 2.0 Flash-Lite 1,136
  113. 113 Llama 4 Scout 1,128
  114. 114 Claude 3.5 Haiku 1,127
  115. 115 Mistral Small 3.1 1,127
  116. 116 Step-1o Vision 32K (High Res) 1,124
  117. 117 Qwen2.5-VL-72B 1,121
  118. 118 Gemini 1.5 Pro May 2024 1,120
  119. 119 Qwen2.5-VL-32B 1,118
  120. 120 GPT-4o Aug 2024 1,118
  121. 121 Molmo2-8B 1,113
  122. 122 GPT-4 Turbo Apr 2024 1,112
  123. 123 GPT-4o mini 1,097
  124. 124 Pixtral Large (2411) 1,095
  125. 125 GPT-4.1 nano 1,089
  126. 126 Qwen2-VL-72B 1,085
  127. 127 Qwen-VL-Max (1119) 1,085
  128. 128 Gemini 1.5 Flash 8B 1,071
  129. 129 Step-1V 32K 1,064
  130. 130 Claude 3 Opus 1,062
  131. 131 Gemini 1.5 Flash May 2024 1,059
  132. 132 Molmo-72B 1,045
  133. 133 Hunyuan Standard Vision 20241231 1,042
  134. 134 Llama 3.2 90B 1,032
  135. 135 Qwen2-VL-7B-Instruct 1,031
  136. 136 Pixtral 12B (2409) 1,025
  137. 137 InternVL2-26B 1,025
  138. 138 Amazon Nova Lite 1,019
  139. 139 Amazon Nova Pro 1,019
  140. 140 Claude 3 Sonnet 1,017
  141. 141 Yi Vision 1,003
  142. 142 Aya Vision 32B 1,001
  143. 143 Claude 3 Haiku 1,000
  144. 144 Molmo-7B-D 994
  145. 145 Llama 3.2 11B 993
  146. 146 nvila-internal-15b-v1 986
  147. 147 LLaVA-OneVision Qwen2 72B OV 978
  148. 148 LLaVA 1.6 34B 965
  149. 149 CogVLM2-19B-Chat 964
  150. 150 MiniCPM-V-2_6 963
  151. 151 InternVL2-4B 959
  152. 152 Phi-3.5-vision-instruct 919
  153. 153 Phi-3-vision-128k-instruct 881

Source: Official Vision Arena leaderboard

#EntryScoreSettingDateSource
1 Claude Fable 5 1,309 high Jun 2026 Official leaderboard
2 Claude Opus 4.7 1,300 Apr 2026 Official leaderboard
3 Qwen 3.8 Max 1,299 Aug 2026 Official leaderboard
4 Claude Opus 4.6 1,299 high Feb 2026 Official leaderboard
5 Claude Opus 5.5 1,295 high Sep 2026 Official leaderboard
6 Muse Spark 1,294 Apr 2026 Official leaderboard
7 Gemini 3.7 Flash 1,294 high Aug 2026 Official leaderboard
8 Muse Spark 1.2 1,294 xhigh Aug 2026 Official leaderboard
9 Gemini 3.8 Flash 1,293 high Sep 2026 Official leaderboard
10 GPT-6.1 Sol 1,293 max Sep 2026 Official leaderboard
11 Muse Spark 1.3 1,292 max Sep 2026 Official leaderboard
12 Gemini 3 Pro 1,289 Nov 2025 Official leaderboard
13 Claude Fable 5.1 1,289 max Sep 2026 Official leaderboard
14 GPT-5.6 Sol 1,288 xhigh Jul 2026 Official leaderboard
15 Claude Opus 4.8 1,286 high May 2026 Official leaderboard
16 Claude Opus 5 1,285 high Jul 2026 Official leaderboard
17 GPT-5.5 1,285 Apr 2026 Official leaderboard
18 GPT-6 Astra 1,283 max Sep 2026 Official leaderboard
19 GPT-5.4 1,282 high Mar 2026 Official leaderboard
20 Gemini 3.5 Flash 1,281 medium May 2026 Official leaderboard
21 Grok 4.5 1,280 Jul 2026 Official leaderboard
22 Gemini 3.1 Pro 1,280 Feb 2026 Official leaderboard
23 Gemini 3.6 Flash 1,280 high Jul 2026 Official leaderboard
24 Muse Spark 1.1 1,279 Jul 2026 Official leaderboard
25 GPT-5.2 Instant 1,278 Dec 2025 Official leaderboard
26 GPT-5.5 Instant 1,277 May 2026 Official leaderboard
27 Claude Sonnet 4.6 1,275 Feb 2026 Official leaderboard
28 Gemini 3 Flash 1,272 Dec 2025 Official leaderboard
29 GLM-5.3-Flash 1,272 Aug 2026 Official leaderboard
30 GPT-5.6 Terra 1,266 xhigh Jul 2026 Official leaderboard
31 Gemini 3.5 Flash-Lite 1,266 Jul 2026 Official leaderboard
32 Gemma 4 31B IT 1,265 Apr 2026 Official leaderboard
33 Kimi K2.6 1,265 Apr 2026 Official leaderboard
34 Grok 4.6 1,263 high Aug 2026 Official leaderboard
35 DeepSeek V4.1 Flash 1,263 max Sep 2026 Official leaderboard
36 Claude Sonnet 5.5 1,263 xhigh Sep 2026 Official leaderboard
37 Qwen3.7-Plus 1,262 Jun 2026 Official leaderboard
38 Claude Sonnet 5 1,261 high Jun 2026 Official leaderboard
39 GPT-6 Sol 1,258 max Sep 2026 Official leaderboard
40 Seed2.0 Pro 1,256 Feb 2026 Official leaderboard
41 Grok 4.20 Beta (Reasoning, 0309) 1,256 Nov 2026 Official leaderboard
42 GPT-5.6 Luna 1,255 xhigh Jul 2026 Official leaderboard
43 Kimi K2.5 1,252 thinking Jan 2026 Official leaderboard
44 GPT-5.1 1,251 high Nov 2025 Official leaderboard
45 Grok 4.20 Multi-Agent Beta (0309) 1,251 Nov 2026 Official leaderboard
46 GPT-5.4 Mini 1,250 high Mar 2026 Official leaderboard
47 Grok 4.7 1,249 xhigh Sep 2026 Official leaderboard
48 MiMo-V2.6 Flash 1,248 Sep 2026 Official leaderboard
49 Gemini 2.5 Pro Jun 2025 1,247 Jun 2025 Official leaderboard
50 MiMo-V2.6 Pro 1,247 Sep 2026 Official leaderboard
51 Qwen3.5 397B-A17B 1,246 Feb 2026 Official leaderboard
52 Step 5 Preview 1,245 Sep 2026 Official leaderboard
53 Qwen 3.8 27B 1,243 Aug 2026 Official leaderboard
54 GPT-5.2 1,242 high Dec 2025 Official leaderboard
55 Gemma 4 26B A4B 1,242 Apr 2026 Official leaderboard
56 Grok 4.3 1,242 Apr 2026 Official leaderboard
57 GPT-4o Mar 2025 1,238 Mar 2025 Official leaderboard
58 GPT-6 Luna 1,235 max Sep 2026 Official leaderboard
59 Gemini 3.1 Flash-Lite 1,235 Mar 2026 Official leaderboard
60 MiMo-V2.5 1,234 Apr 2026 Official leaderboard
61 MiniMax-M3 1,232 Jun 2026 Official leaderboard
62 GLM-5V-Turbo 1,232 Apr 2026 Official leaderboard
63 Mistral Large 4 1,231 Oct 2026 Official leaderboard
64 Gemini 2.5 Flash Sep 2025 1,229 Sep 2025 Official leaderboard
65 Qwen3.5-122B-A10B 1,226 Feb 2026 Official leaderboard
66 GPT-5 Chat 1,225 Aug 2025 Official leaderboard
67 GPT-4.5 1,225 Feb 2025 Official leaderboard
68 ERNIE-5.0-Preview (1220) 1,219 Nov 2026 Official leaderboard
69 MiMo-V2-Omni 1,216 Mar 2026 Official leaderboard
70 Qwen3.5-27B 1,216 Feb 2026 Official leaderboard
71 o3 1,216 Apr 2025 Official leaderboard
72 Gemini 2.5 Flash Jun 2025 1,215 Jun 2025 Official leaderboard
73 Qwen3-VL-235B-A22B-Instruct 1,214 Sep 2025 Official leaderboard
74 GPT-4.1 1,214 Apr 2025 Official leaderboard
75 Nemotron 3.5 Super VL 1,212 nvfp4 Oct 2026 Official leaderboard
76 GPT-5 1,211 high Aug 2025 Official leaderboard
77 Claude Opus 4 1,208 thinking 16k May 2025 Official leaderboard
78 Inkling-Small 1,208 Jul 2026 Official leaderboard
79 Claude Sonnet 4 1,206 thinking 32k May 2025 Official leaderboard
80 o4-mini 1,201 Apr 2025 Official leaderboard
81 GPT-4.1 mini 1,201 Apr 2025 Official leaderboard
82 Mistral Large 3 1,200 Dec 2025 Official leaderboard
83 Mistral Medium 3.5 1,199 Apr 2026 Official leaderboard
84 GPT-5.4 Nano 1,199 high Mar 2026 Official leaderboard
85 Grok 4.1 Fast 1,196 reasoning Nov 2025 Official leaderboard
86 o1 1,193 Dec 2024 Official leaderboard
87 Claude 3.7 Sonnet 1,193 thinking 32k Feb 2025 Official leaderboard
88 Gemini 2.5 Flash-Lite Jun 2025 1,189 Jun 2025 Official leaderboard
89 Qwen3-VL-235B-A22B 1,188 thinking Sep 2025 Official leaderboard
90 GPT-5 mini 1,185 high Aug 2025 Official leaderboard
91 Qwen-VL-Max 1,185 Jan 2024 Official leaderboard
92 Grok 4 1,184 Jul 2025 Official leaderboard
93 Gemini 1.5 Pro Sep 2024 1,180 Sep 2024 Official leaderboard
94 Gemini 2.5 Flash-Lite Sep 2025 1,174 non-thinking, 09-2025 Sep 2025 Official leaderboard
95 Gemini 2.0 Flash Feb 2025 1,172 Feb 2025 Official leaderboard
96 GPT-4o May 2024 1,162 May 2024 Official leaderboard
97 Claude 3.5 Sonnet (October 2024) 1,161 Oct 2024 Official leaderboard
98 GLM-4.6V 1,161 Dec 2025 Official leaderboard
99 Gemma 3 27B 1,160 Mar 2025 Official leaderboard
100 Mistral Medium 3.1 1,160 Aug 2025 Official leaderboard
101 Step-1o Turbo (202506) 1,155 Nov 2026 Official leaderboard
102 Mistral Medium 3 1,155 May 2025 Official leaderboard
103 Hunyuan Vision 1.5 Thinking 1,154 Nov 2026 Official leaderboard
104 GLM-4.5V 1,153 Aug 2025 Official leaderboard
105 Hunyuan Large Vision 1,150 Nov 2026 Official leaderboard
106 GPT-5 nano 1,147 high Aug 2025 Official leaderboard
107 Claude 3.5 Sonnet 1,146 Jun 2024 Official leaderboard
108 Llama 4 Maverick 1,145 Apr 2025 Official leaderboard
109 Step 3 1,144 Jul 2025 Official leaderboard
110 Gemini 1.5 Flash Sep 2024 1,142 Aug 2024 Official leaderboard
111 Mistral Small 3.2 1,140 Jun 2025 Official leaderboard
112 Gemini 2.0 Flash-Lite 1,136 Feb 2025 Official leaderboard
113 Llama 4 Scout 1,128 Apr 2025 Official leaderboard
114 Claude 3.5 Haiku 1,127 Oct 2024 Official leaderboard
115 Mistral Small 3.1 1,127 Mar 2025 Official leaderboard
116 Step-1o Vision 32K (High Res) 1,124 Nov 2026 Official leaderboard
117 Qwen2.5-VL-72B 1,121 Jan 2025 Official leaderboard
118 Gemini 1.5 Pro May 2024 1,120 May 2024 Official leaderboard
119 Qwen2.5-VL-32B 1,118 Mar 2025 Official leaderboard
120 GPT-4o Aug 2024 1,118 Aug 2024 Official leaderboard
121 Molmo2-8B 1,113 Dec 2025 Official leaderboard
122 GPT-4 Turbo Apr 2024 1,112 Apr 2024 Official leaderboard
123 GPT-4o mini 1,097 Jul 2024 Official leaderboard
124 Pixtral Large (2411) 1,095 Nov 2026 Official leaderboard
125 GPT-4.1 nano 1,089 Apr 2025 Official leaderboard
126 Qwen2-VL-72B 1,085 Nov 2026 Official leaderboard
127 Qwen-VL-Max (1119) 1,085 Nov 2026 Official leaderboard
128 Gemini 1.5 Flash 8B 1,071 Oct 2024 Official leaderboard
129 Step-1V 32K 1,064 Nov 2026 Official leaderboard
130 Claude 3 Opus 1,062 Feb 2024 Official leaderboard
131 Gemini 1.5 Flash May 2024 1,059 May 2024 Official leaderboard
132 Molmo-72B 1,045 Sep 2024 Official leaderboard
133 Hunyuan Standard Vision 20241231 1,042 Dec 2024 Official leaderboard
134 Llama 3.2 90B 1,032 Sep 2024 Official leaderboard
135 Qwen2-VL-7B-Instruct 1,031 Aug 2024 Official leaderboard
136 Pixtral 12B (2409) 1,025 Nov 2026 Official leaderboard
137 InternVL2-26B 1,025 Jul 2024 Official leaderboard
138 Amazon Nova Lite 1,019 Dec 2024 Official leaderboard
139 Amazon Nova Pro 1,019 Dec 2024 Official leaderboard
140 Claude 3 Sonnet 1,017 Feb 2024 Official leaderboard
141 Yi Vision 1,003 Nov 2026 Official leaderboard
142 Aya Vision 32B 1,001 Mar 2025 Official leaderboard
143 Claude 3 Haiku 1,000 Mar 2024 Official leaderboard
144 Molmo-7B-D 994 Sep 2024 Official leaderboard
145 Llama 3.2 11B 993 Sep 2024 Official leaderboard
146 nvila-internal-15b-v1 986 Nov 2026 Official leaderboard
147 LLaVA-OneVision Qwen2 72B OV 978 Nov 2026 Official leaderboard
148 LLaVA 1.6 34B 965 Nov 2026 Official leaderboard
149 CogVLM2-19B-Chat 964 May 2024 Official leaderboard
150 MiniCPM-V-2_6 963 Aug 2024 Official leaderboard
151 InternVL2-4B 959 Jul 2024 Official leaderboard
152 Phi-3.5-vision-instruct 919 Aug 2024 Official leaderboard
153 Phi-3-vision-128k-instruct 881 May 2024 Official leaderboard

More effort adds a median 4 elo

Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.

Lowest effort to highest, light to dark
Claude Opus 4.7
high
1,300
Claude Opus 4.6
high
1,299
Muse Spark 1.3
xhigh
1,292
Claude Opus 4.8
high
1,286
GPT-5.5
high
1,285
GPT-5.4
high
1,282
Gemini 3.5 Flash
medium
1,281
Gemini 3 Flash
minimal
1,272
Kimi K2.5
instant thinking
1,252
GPT-5.1
high
1,251
GPT-5.2
high
1,242
Claude Opus 4
thinking 16k
1,208

Source: Official Vision Arena leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Claude Fable 5 high 1,309
Claude Opus 4.7 1,300
Qwen 3.8 Max 1,299
Claude Opus 4.6 high 1,299
Claude Opus 4.7 high 1,298
Claude Opus 5.5 high 1,295
Muse Spark 1,294
Gemini 3.7 Flash high 1,294
Muse Spark 1.2 xhigh 1,294
Claude Opus 4.6 1,293
Gemini 3.8 Flash high 1,293
GPT-6.1 Sol max 1,293
Muse Spark 1.3 max 1,292
Gemini 3 Pro 1,289
Claude Fable 5.1 max 1,289
GPT-5.6 Sol xhigh 1,288
Muse Spark 1.3 xhigh 1,288
Claude Opus 4.8 high 1,286
Claude Opus 5 high 1,285
GPT-5.5 1,285
GPT-6 Astra max 1,283
GPT-5.5 high 1,283
GPT-5.4 high 1,282
Gemini 3.5 Flash medium 1,281
Gemini 3.5 Flash high 1,281
Grok 4.5 1,280
Gemini 3.1 Pro 1,280
Gemini 3.6 Flash high 1,280
Muse Spark 1.1 1,279
Claude Opus 4.8 1,279
GPT-5.4 1,278
GPT-5.2 Instant 1,278
GPT-5.5 Instant 1,277
Claude Sonnet 4.6 1,275
Gemini 3 Flash 1,272
GLM-5.3-Flash 1,272
GPT-5.6 Terra xhigh 1,266
Gemini 3.5 Flash-Lite 1,266
Gemma 4 31B IT 1,265
Kimi K2.6 1,265
Grok 4.6 high 1,263
DeepSeek V4.1 Flash max 1,263
Claude Sonnet 5.5 xhigh 1,263
Qwen3.7-Plus 1,262
Claude Sonnet 5 high 1,261
Gemini 3 Flash minimal 1,260
GPT-6 Sol max 1,258
Seed2.0 Pro 1,256
GPT-5.6 Luna xhigh 1,255
Kimi K2.5 thinking 1,252
GPT-5.1 high 1,251
GPT-5.4 Mini high 1,250
Grok 4.7 xhigh 1,249
MiMo-V2.6 Flash 1,248
Gemini 2.5 Pro Jun 2025 1,247
MiMo-V2.6 Pro 1,247
Qwen3.5 397B-A17B 1,246
Step 5 Preview 1,245
Qwen 3.8 27B 1,243
GPT-5.2 high 1,242
Gemma 4 26B A4B 1,242
Grok 4.3 1,242
GPT-5.1 1,240
GPT-4o Mar 2025 1,238
Kimi K2.5 instant 1,236
GPT-6 Luna max 1,235
Gemini 3.1 Flash-Lite 1,235
MiMo-V2.5 1,234
MiniMax-M3 1,232
GLM-5V-Turbo 1,232
Mistral Large 4 1,231
Gemini 2.5 Flash Sep 2025 1,229
GPT-5.2 1,228
Qwen3.5-122B-A10B 1,226
GPT-5 Chat 1,225
GPT-4.5 1,225
MiMo-V2-Omni 1,216
Qwen3.5-27B 1,216
o3 1,216
Gemini 2.5 Flash Jun 2025 1,215
Qwen3-VL-235B-A22B-Instruct 1,214
GPT-4.1 1,214
Nemotron 3.5 Super VL nvfp4 1,212
GPT-5 high 1,211
Claude Opus 4 thinking 16k 1,208
Inkling-Small 1,208
Claude Sonnet 4 thinking 32k 1,206
o4-mini 1,201
GPT-4.1 mini 1,201
Mistral Large 3 1,200
Mistral Medium 3.5 1,199
GPT-5.4 Nano high 1,199
Grok 4.1 Fast reasoning 1,196
o1 1,193
Claude 3.7 Sonnet thinking 32k 1,193
Claude Opus 4 1,189
Gemini 2.5 Flash-Lite Jun 2025 1,189
Claude Sonnet 4 1,189
Qwen3-VL-235B-A22B thinking 1,188
GPT-5 mini high 1,185
Qwen-VL-Max 1,185
Grok 4 1,184
Gemini 1.5 Pro Sep 2024 1,180
Claude 3.7 Sonnet 1,175
Gemini 2.5 Flash-Lite Sep 2025 non-thinking, 09-2025 1,174
Gemini 2.0 Flash Feb 2025 1,172
GPT-4o May 2024 1,162
Claude 3.5 Sonnet (October 2024) 1,161
GLM-4.6V 1,161
Gemma 3 27B 1,160
Mistral Medium 3.1 1,160
Mistral Medium 3 1,155
GLM-4.5V 1,153
GPT-5 nano high 1,147
Claude 3.5 Sonnet 1,146
Llama 4 Maverick 1,145
Step 3 1,144
Gemini 1.5 Flash Sep 2024 1,142
Mistral Small 3.2 1,140
Gemini 2.0 Flash-Lite 1,136
Llama 4 Scout 1,128
Claude 3.5 Haiku 1,127
Mistral Small 3.1 1,127
Qwen2.5-VL-72B 1,121
Gemini 1.5 Pro May 2024 1,120
Qwen2.5-VL-32B 1,118
GPT-4o Aug 2024 1,118
Molmo2-8B 1,113
GPT-4 Turbo Apr 2024 1,112
GPT-4o mini 1,097
GPT-4.1 nano 1,089
Gemini 1.5 Flash 8B 1,071
Claude 3 Opus 1,062
Gemini 1.5 Flash May 2024 1,059
Molmo-72B 1,045
Llama 3.2 90B 1,032
Qwen2-VL-7B-Instruct 1,031
InternVL2-26B 1,025
Amazon Nova Lite 1,019
Amazon Nova Pro 1,019
Claude 3 Sonnet 1,017
Aya Vision 32B 1,001
Claude 3 Haiku 1,000
Molmo-7B-D 994
Llama 3.2 11B 993
CogVLM2-19B-Chat 964
MiniCPM-V-2_6 963
InternVL2-4B 959
Phi-3.5-vision-instruct 919
Phi-3-vision-128k-instruct 881

10 models have held number one on the official leaderboard since its release in June 2024

Lab claims count from the day the card came out, independent runs from the model’s release.

Other labs OpenAI Google DeepMind Anthropic
Own leaderboard 10 leaders
Claude Fable 5 1,309

Source: Official Vision Arena leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

On the benchmark’s own leaderboard
800 1,000 1,200 1,400 2024 2025 2026 800 1,000 1,200 1,400 2024 2025 2026

Source: Official Vision Arena leaderboard

No lab model card reports it yet

We read the headline results of 522 launch posts from 52 labs, and none of them list this benchmark.

More in Vision

All 281 in Vision
  • MMMU In use 11.5K college-level multimodal questions across 30 subjects. Multiple-choice and open accuracy. GPT-5.1 85.4%
  • 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%