Acropolis

LMArena Vision

Arena human-vote leaderboard: rating with confidence interval, vote count and rank. One row per model and category.

Vision
Independent. Run by an operator independent of the model vendors.LMArena by ArenaVersion 1Data as of 17 Nov 2026

No filters. Showing 167 of 167 rows.

  1. 1
    AnthropicDossier

    high

    Arena rating

    Fable 5, Arena rating: 1,326 (scale 750.0 to 1325.6), 95% confidence interval 1318.7 to 1332.51,326(1318.7–1332.5)
  2. 2
    AnthropicDossier

    high

    Tied #1

    Arena rating

    Opus 5.5, Arena rating: 1,324 (scale 750.0 to 1325.6), 95% confidence interval 1314.0 to 1333.61,324(1314.0–1333.6)
  3. 3
    AnthropicDossier

    max

    Tied #1

    Arena rating

    Fable 5.1, Arena rating: 1,322 (scale 750.0 to 1325.6), 95% confidence interval 1313.0 to 1330.21,322(1313.0–1330.2)
  4. 4
    AnthropicDossier
    Tied #1

    Arena rating

    Opus 4.7, Arena rating: 1,319 (scale 750.0 to 1325.6), 95% confidence interval 1312.2 to 1325.31,319(1312.2–1325.3)
  5. 5
    AnthropicDossier

    high

    Tied #1

    Arena rating

    Opus 5, Arena rating: 1,317 (scale 750.0 to 1325.6), 95% confidence interval 1311.0 to 1323.91,317(1311.0–1323.9)
  6. 6
    GoogleDossier

    high

    Tied #1

    Arena rating

    Gemini 3.8 Flash, Arena rating: 1,317 (scale 750.0 to 1325.6), 95% confidence interval 1308.6 to 1324.91,317(1308.6–1324.9)
  7. 7
    AnthropicDossier

    high

    Arena rating

    Opus 4.6, Arena rating: 1,316 (scale 750.0 to 1325.6), 95% confidence interval 1309.2 to 1321.91,316(1309.2–1321.9)
  8. 8
    AnthropicDossier

    high

    Arena rating

    Opus 4.7, Arena rating: 1,314 (scale 750.0 to 1325.6), 95% confidence interval 1307.4 to 1320.51,314(1307.4–1320.5)
  9. 9
    GoogleDossier

    high

    Arena rating

    Gemini 3.7 Flash, Arena rating: 1,313 (scale 750.0 to 1325.6), 95% confidence interval 1304.7 to 1321.61,313(1304.7–1321.6)
  10. 10
    AnthropicDossier

    Arena rating

    Opus 4.6, Arena rating: 1,312 (scale 750.0 to 1325.6), 95% confidence interval 1306.0 to 1318.41,312(1306.0–1318.4)
  11. 11
    AlibabaDossier

    Arena rating

    Qwen3.8-Max, Arena rating: 1,312 (scale 750.0 to 1325.6), 95% confidence interval 1304.6 to 1319.41,312(1304.6–1319.4)
  12. 12

    max

    Arena rating

    Muse Spark 1.3, Arena rating: 1,310 (scale 750.0 to 1325.6), 95% confidence interval 1301.5 to 1317.61,310(1301.5–1317.6)
  13. 13
    GoogleDossier

    high

    Arena rating

    Gemini 3.5 Flash, Arena rating: 1,308 (scale 750.0 to 1325.6), 95% confidence interval 1300.9 to 1314.11,308(1300.9–1314.1)
  14. 14

    Arena rating

    Muse Spark, Arena rating: 1,306 (scale 750.0 to 1325.6), 95% confidence interval 1297.0 to 1315.01,306(1297.0–1315.0)
  15. 15
    GoogleDossier

    medium

    Arena rating

    Gemini 3.5 Flash, Arena rating: 1,305 (scale 750.0 to 1325.6), 95% confidence interval 1298.9 to 1312.11,305(1298.9–1312.1)
  16. 16

    xhigh

    Arena rating

    Muse Spark 1.2, Arena rating: 1,305 (scale 750.0 to 1325.6), 95% confidence interval 1290.6 to 1319.31,305(1290.6–1319.3)
  17. 17
    GoogleDossier

    Arena rating

    Gemini 3 Pro Preview, Arena rating: 1,305 (scale 750.0 to 1325.6), 95% confidence interval 1297.6 to 1312.31,305(1297.6–1312.3)
  18. 18

    xhigh

    Arena rating

    Muse Spark 1.3, Arena rating: 1,301 (scale 750.0 to 1325.6), 95% confidence interval 1292.6 to 1309.01,301(1292.6–1309.0)
  19. 19
    OpenAIDossier

    high

    Arena rating

    GPT-5.4, Arena rating: 1,301 (scale 750.0 to 1325.6), 95% confidence interval 1295.2 to 1306.31,301(1295.2–1306.3)
  20. 20
    GoogleDossier

    high

    Arena rating

    Gemini 3.6 Flash, Arena rating: 1,297 (scale 750.0 to 1325.6), 95% confidence interval 1289.2 to 1304.21,297(1289.2–1304.2)
  21. 21
    GoogleDossier

    Arena rating

    Gemini 3.1 Pro Preview, Arena rating: 1,296 (scale 750.0 to 1325.6), 95% confidence interval 1291.0 to 1301.01,296(1291.0–1301.0)
  22. 22
    OpenAIDossier

    Arena rating

    GPT-5.5, Arena rating: 1,296 (scale 750.0 to 1325.6), 95% confidence interval 1290.2 to 1301.51,296(1290.2–1301.5)
  23. 23
    AnthropicDossier

    high

    Arena rating

    Opus 4.8, Arena rating: 1,295 (scale 750.0 to 1325.6), 95% confidence interval 1288.3 to 1301.11,295(1288.3–1301.1)
  24. 24

    Arena rating

    GLM-5.3 Flash, Arena rating: 1,295 (scale 750.0 to 1325.6), 95% confidence interval 1289.2 to 1300.21,295(1289.2–1300.2)
  25. 25
    OpenAIDossier

    high

    Arena rating

    GPT-5.5, Arena rating: 1,295 (scale 750.0 to 1325.6), 95% confidence interval 1288.8 to 1300.21,295(1288.8–1300.2)
  26. 26

    Arena rating

    Muse Spark 1.1, Arena rating: 1,292 (scale 750.0 to 1325.6), 95% confidence interval 1284.9 to 1298.51,292(1284.9–1298.5)
  27. 27
    OpenAIDossier

    Arena rating

    GPT-5.4, Arena rating: 1,292 (scale 750.0 to 1325.6), 95% confidence interval 1285.3 to 1298.11,292(1285.3–1298.1)
  28. 28
    AnthropicDossier

    Arena rating

    Opus 4.8, Arena rating: 1,289 (scale 750.0 to 1325.6), 95% confidence interval 1282.2 to 1295.11,289(1282.2–1295.1)
  29. 29

    Arena rating

    Grok 4.5, Arena rating: 1,288 (scale 750.0 to 1325.6), 95% confidence interval 1281.2 to 1294.81,288(1281.2–1294.8)
  30. 30
    OpenAIDossier

    max

    Arena rating

    GPT-6.1 Sol, Arena rating: 1,286 (scale 750.0 to 1325.6), 95% confidence interval 1275.5 to 1296.41,286(1275.5–1296.4)
  31. 31
    GoogleDossier

    Arena rating

    Gemini 3 Flash Preview, Arena rating: 1,285 (scale 750.0 to 1325.6), 95% confidence interval 1279.5 to 1289.81,285(1279.5–1289.8)
  32. 32
    AnthropicDossier

    xhigh

    Arena rating

    Sonnet 5.5, Arena rating: 1,284 (scale 750.0 to 1325.6), 95% confidence interval 1274.1 to 1294.41,284(1274.1–1294.4)
  33. 33
    AnthropicDossier

    Arena rating

    Sonnet 4.6, Arena rating: 1,283 (scale 750.0 to 1325.6), 95% confidence interval 1276.6 to 1288.71,283(1276.6–1288.7)
  34. 34
    Moonshot AIDossier

    Arena rating

    Kimi K2.6, Arena rating: 1,282 (scale 750.0 to 1325.6), 95% confidence interval 1275.7 to 1289.21,282(1275.7–1289.2)
  35. 35
    OpenAIDossier

    xhigh

    Arena rating

    GPT-5.6 Sol, Arena rating: 1,282 (scale 750.0 to 1325.6), 95% confidence interval 1275.1 to 1289.31,282(1275.1–1289.3)
  36. 36
    GoogleDossier

    Arena rating

    Gemma 4 31B, Arena rating: 1,279 (scale 750.0 to 1325.6), 95% confidence interval 1273.9 to 1284.91,279(1273.9–1284.9)
  37. 37
    DeepSeekDossier

    max

    Arena rating

    DeepSeek V4.1 Flash, Arena rating: 1,279 (scale 750.0 to 1325.6), 95% confidence interval 1269.1 to 1289.21,279(1269.1–1289.2)
  38. 38
    OpenAIDossier

    max

    Arena rating

    GPT-6 Astra, Arena rating: 1,279 (scale 750.0 to 1325.6), 95% confidence interval 1269.4 to 1287.61,279(1269.4–1287.6)
  39. 39
    AlibabaDossier

    Arena rating

    Qwen3.7 Plus, Arena rating: 1,277 (scale 750.0 to 1325.6), 95% confidence interval 1270.7 to 1283.81,277(1270.7–1283.8)
  40. 40
    ByteDanceDossier

    Arena rating

    Dola Seed 2.0 Pro, Arena rating: 1,274 (scale 750.0 to 1325.6), 95% confidence interval 1267.0 to 1281.31,274(1267.0–1281.3)
  41. 41
    AnthropicDossier

    high

    Arena rating

    Sonnet 5, Arena rating: 1,272 (scale 750.0 to 1325.6), 95% confidence interval 1264.9 to 1278.31,272(1264.9–1278.3)
  42. 42
    AlibabaDossier

    Arena rating

    Qwen3.8-27B, Arena rating: 1,270 (scale 750.0 to 1325.6), 95% confidence interval 1262.1 to 1277.21,270(1262.1–1277.2)
  43. 43
    OpenAIDossier

    xhigh

    Arena rating

    GPT-5.6 Terra, Arena rating: 1,269 (scale 750.0 to 1325.6), 95% confidence interval 1262.1 to 1276.11,269(1262.1–1276.1)
  44. 44
    Moonshot AIDossier

    thinking

    Arena rating

    Kimi K2.5, Arena rating: 1,269 (scale 750.0 to 1325.6), 95% confidence interval 1263.2 to 1274.51,269(1263.2–1274.5)
  45. 45
    GoogleDossier

    Arena rating

    Gemini 3.5 Flash-Lite, Arena rating: 1,268 (scale 750.0 to 1325.6), 95% confidence interval 1260.5 to 1275.41,268(1260.5–1275.4)
  46. 46
    OpenAIDossier

    Arena rating

    GPT-5.2 Chat, Arena rating: 1,268 (scale 750.0 to 1325.6), 95% confidence interval 1261.3 to 1274.11,268(1261.3–1274.1)
  47. 47
    XiaomiDossier

    Arena rating

    MiMo-V2.6-Pro, Arena rating: 1,266 (scale 750.0 to 1325.6), 95% confidence interval 1251.4 to 1280.41,266(1251.4–1280.4)
  48. 48

    Arena rating

    GLM-5V-Turbo, Arena rating: 1,266 (scale 750.0 to 1325.6), 95% confidence interval 1260.4 to 1271.01,266(1260.4–1271.0)
  49. 49
    GoogleDossier

    minimal

    Arena rating

    Gemini 3 Flash Preview, Arena rating: 1,265 (scale 750.0 to 1325.6), 95% confidence interval 1260.1 to 1270.61,265(1260.1–1270.6)
  50. 50

    high

    Arena rating

    Grok 4.6, Arena rating: 1,264 (scale 750.0 to 1325.6), 95% confidence interval 1256.1 to 1271.91,264(1256.1–1271.9)
  51. 51
    GoogleDossier

    Arena rating

    Gemini 2.5 Pro, Arena rating: 1,263 (scale 750.0 to 1325.6), 95% confidence interval 1258.7 to 1267.71,263(1258.7–1267.7)
  52. 52
    StepFunDossier

    Arena rating

    Step 5 Preview, Arena rating: 1,263 (scale 750.0 to 1325.6), 95% confidence interval 1252.3 to 1274.11,263(1252.3–1274.1)
  53. 53

    thinking

    Arena rating

    Grok 4.20 0309, Arena rating: 1,263 (scale 750.0 to 1325.6), 95% confidence interval 1256.9 to 1268.91,263(1256.9–1268.9)
  54. 54
    AlibabaDossier

    Arena rating

    Qwen3.5 397B A17B, Arena rating: 1,262 (scale 750.0 to 1325.6), 95% confidence interval 1257.2 to 1267.61,262(1257.2–1267.6)
  55. 55
    GoogleDossier

    Arena rating

    Gemma 4 26B A4B, Arena rating: 1,260 (scale 750.0 to 1325.6), 95% confidence interval 1253.3 to 1266.11,260(1253.3–1266.1)
  56. 56
    XiaomiDossier

    Arena rating

    MiMo-V2.6-Flash, Arena rating: 1,260 (scale 750.0 to 1325.6), 95% confidence interval 1251.0 to 1268.11,260(1251.0–1268.1)
  57. 57

    Arena rating

    Grok 4.20 Multi-Agent Beta (0309), Arena rating: 1,259 (scale 750.0 to 1325.6), 95% confidence interval 1253.1 to 1265.21,259(1253.1–1265.2)
  58. 58
    OpenAIDossier

    xhigh

    Arena rating

    GPT-5.6 Luna, Arena rating: 1,256 (scale 750.0 to 1325.6), 95% confidence interval 1248.5 to 1263.21,256(1248.5–1263.2)
  59. 59
    GoogleDossier

    Arena rating

    Gemini 2.5 Flash Preview (09-2025), Arena rating: 1,253 (scale 750.0 to 1325.6), 95% confidence interval 1243.4 to 1262.81,253(1243.4–1262.8)
  60. 60
    Moonshot AIDossier

    instant

    Arena rating

    Kimi K2.5, Arena rating: 1,253 (scale 750.0 to 1325.6), 95% confidence interval 1242.3 to 1263.61,253(1242.3–1263.6)
  61. 61
    OpenAIDossier

    Arena rating

    GPT-5.5 Instant, Arena rating: 1,251 (scale 750.0 to 1325.6), 95% confidence interval 1242.0 to 1259.11,251(1242.0–1259.1)
  62. 62
    MiniMaxDossier

    Arena rating

    MiniMax M3, Arena rating: 1,249 (scale 750.0 to 1325.6), 95% confidence interval 1243.0 to 1255.41,249(1243.0–1255.4)
  63. 63
    OpenAIDossier

    high

    Arena rating

    GPT-5.1, Arena rating: 1,249 (scale 750.0 to 1325.6), 95% confidence interval 1241.3 to 1256.91,249(1241.3–1256.9)
  64. 64
    BaiduDossier

    Arena rating

    ERNIE 5.0 Preview (1220), Arena rating: 1,248 (scale 750.0 to 1325.6), 95% confidence interval 1237.1 to 1259.31,248(1237.1–1259.3)
  65. 65
    XiaomiDossier

    Arena rating

    MiMo-V2.5, Arena rating: 1,248 (scale 750.0 to 1325.6), 95% confidence interval 1241.7 to 1253.31,248(1241.7–1253.3)
  66. 66
    AlibabaDossier

    Arena rating

    Qwen3 VL 235B A22B Instruct, Arena rating: 1,247 (scale 750.0 to 1325.6), 95% confidence interval 1239.7 to 1253.51,247(1239.7–1253.5)
  67. 67
    OpenAIDossier

    high

    Arena rating

    GPT-5.2, Arena rating: 1,245 (scale 750.0 to 1325.6), 95% confidence interval 1239.5 to 1251.51,245(1239.5–1251.5)
  68. 68
    OpenAIDossier

    high

    Arena rating

    GPT-5.4 mini, Arena rating: 1,244 (scale 750.0 to 1325.6), 95% confidence interval 1238.3 to 1250.51,244(1238.3–1250.5)
  69. 69
    AlibabaDossier

    Arena rating

    Qwen3.5 122B A10B, Arena rating: 1,244 (scale 750.0 to 1325.6), 95% confidence interval 1237.5 to 1250.91,244(1237.5–1250.9)
  70. 70
    MistralDossier

    Arena rating

    Mistral Large 4, Arena rating: 1,244 (scale 750.0 to 1325.6), 95% confidence interval 1229.9 to 1258.11,244(1229.9–1258.1)
  71. 71
    OpenAIDossier

    Arena rating

    ChatGPT-4o Latest, Arena rating: 1,243 (scale 750.0 to 1325.6), 95% confidence interval 1237.8 to 1248.71,243(1237.8–1248.7)
  72. 72
    GoogleDossier

    Arena rating

    Gemini 3.1 Flash-Lite Preview, Arena rating: 1,240 (scale 750.0 to 1325.6), 95% confidence interval 1234.6 to 1246.01,240(1234.6–1246.0)
  73. 73
    AlibabaDossier

    Arena rating

    Qwen3.5 27B, Arena rating: 1,240 (scale 750.0 to 1325.6), 95% confidence interval 1234.3 to 1245.01,240(1234.3–1245.0)
  74. 74
    Thinking MachinesDossier

    Arena rating

    Inkling Small, Arena rating: 1,236 (scale 750.0 to 1325.6), 95% confidence interval 1228.1 to 1243.71,236(1228.1–1243.7)
  75. 75
    GoogleDossier

    Arena rating

    Gemini 2.5 Flash, Arena rating: 1,235 (scale 750.0 to 1325.6), 95% confidence interval 1230.8 to 1239.81,235(1230.8–1239.8)
  76. 76
    OpenAIDossier

    Arena rating

    GPT-5.1, Arena rating: 1,235 (scale 750.0 to 1325.6), 95% confidence interval 1227.5 to 1242.51,235(1227.5–1242.5)
  77. 77
    OpenAIDossier

    max

    Arena rating

    GPT-6 Sol, Arena rating: 1,233 (scale 750.0 to 1325.6), 95% confidence interval 1223.0 to 1242.31,233(1223.0–1242.3)
  78. 78
    OpenAIDossier

    Arena rating

    GPT-5 Chat, Arena rating: 1,232 (scale 750.0 to 1325.6), 95% confidence interval 1224.6 to 1238.71,232(1224.6–1238.7)
  79. 79

    xhigh

    Arena rating

    Grok 4.7, Arena rating: 1,230 (scale 750.0 to 1325.6), 95% confidence interval 1221.2 to 1238.91,230(1221.2–1238.9)
  80. 80
    OpenAIDossier

    Arena rating

    GPT-5.2, Arena rating: 1,229 (scale 750.0 to 1325.6), 95% confidence interval 1222.9 to 1234.81,229(1222.9–1234.8)
  81. 81

    Arena rating

    Grok 4.3, Arena rating: 1,229 (scale 750.0 to 1325.6), 95% confidence interval 1222.9 to 1234.31,229(1222.9–1234.3)
  82. 82
    XiaomiDossier

    Arena rating

    MiMo-V2-Omni, Arena rating: 1,228 (scale 750.0 to 1325.6), 95% confidence interval 1220.7 to 1235.31,228(1220.7–1235.3)
  83. 83
    MistralDossier

    Arena rating

    Mistral Medium 3.5, Arena rating: 1,222 (scale 750.0 to 1325.6), 95% confidence interval 1212.9 to 1231.91,222(1212.9–1231.9)
  84. 84
    MistralDossier

    Arena rating

    Mistral Large 3, Arena rating: 1,222 (scale 750.0 to 1325.6), 95% confidence interval 1214.3 to 1229.31,222(1214.3–1229.3)
  85. 85
    NVIDIADossier

    Arena rating

    Nemotron 3.5 Super VL 120B A12B (NVFP4), Arena rating: 1,214 (scale 750.0 to 1325.6), 95% confidence interval 1205.1 to 1223.71,214(1205.1–1223.7)
  86. 86
    OpenAIDossier

    Arena rating

    o3, Arena rating: 1,214 (scale 750.0 to 1325.6), 95% confidence interval 1207.2 to 1220.31,214(1207.2–1220.3)
  87. 87
    AlibabaDossier

    Arena rating

    Qwen VL Max, Arena rating: 1,212 (scale 750.0 to 1325.6), 95% confidence interval 1200.4 to 1224.31,212(1200.4–1224.3)
  88. 88
    OpenAIDossier

    max

    Arena rating

    GPT-6 Luna, Arena rating: 1,211 (scale 750.0 to 1325.6), 95% confidence interval 1202.3 to 1220.11,211(1202.3–1220.1)
  89. 89

    Arena rating

    Grok 4 (0709), Arena rating: 1,210 (scale 750.0 to 1325.6), 95% confidence interval 1203.0 to 1217.61,210(1203.0–1217.6)
  90. 90
    OpenAIDossier

    Arena rating

    GPT-4.1, Arena rating: 1,210 (scale 750.0 to 1325.6), 95% confidence interval 1203.5 to 1216.91,210(1203.5–1216.9)
  91. 91
    OpenAIDossier

    high

    Arena rating

    GPT-5, Arena rating: 1,208 (scale 750.0 to 1325.6), 95% confidence interval 1201.2 to 1214.81,208(1201.2–1214.8)
  92. 92
    AlibabaDossier

    Arena rating

    Qwen3 VL 235B A22B Thinking, Arena rating: 1,206 (scale 750.0 to 1325.6), 95% confidence interval 1194.0 to 1218.51,206(1194.0–1218.5)
  93. 93
    OpenAIDossier

    high

    Arena rating

    GPT-5 mini, Arena rating: 1,202 (scale 750.0 to 1325.6), 95% confidence interval 1194.7 to 1210.31,202(1194.7–1210.3)
  94. 94

    thinking

    Arena rating

    Grok 4.1 Fast, Arena rating: 1,201 (scale 750.0 to 1325.6), 95% confidence interval 1194.4 to 1208.21,201(1194.4–1208.2)
  95. 95
    GoogleDossier

    non-thinking

    Arena rating

    Gemini 2.5 Flash-Lite Preview (Sep '25), Arena rating: 1,198 (scale 750.0 to 1325.6), 95% confidence interval 1188.6 to 1207.61,198(1188.6–1207.6)
  96. 96
    OpenAIDossier

    high

    Arena rating

    GPT-5.4 nano, Arena rating: 1,196 (scale 750.0 to 1325.6), 95% confidence interval 1189.3 to 1201.81,196(1189.3–1201.8)
  97. 97
    OpenAIDossier

    Arena rating

    GPT-4.5 Preview, Arena rating: 1,195 (scale 750.0 to 1325.6), 95% confidence interval 1183.0 to 1206.51,195(1183.0–1206.5)
  98. 98
    OpenAIDossier

    Arena rating

    o4-mini, Arena rating: 1,194 (scale 750.0 to 1325.6), 95% confidence interval 1186.9 to 1201.01,194(1186.9–1201.0)
  99. 99
    AnthropicDossier

    thinking · 16,000 budget

    Arena rating

    Opus 4, Arena rating: 1,192 (scale 750.0 to 1325.6), 95% confidence interval 1175.9 to 1207.91,192(1175.9–1207.9)
  100. 100
    AnthropicDossier

    thinking · 32,000 budget

    Arena rating

    Claude Sonnet 4, Arena rating: 1,191 (scale 750.0 to 1325.6), 95% confidence interval 1173.8 to 1207.81,191(1173.8–1207.8)
  101. 101
    GoogleDossier

    thinking

    Arena rating

    Gemini 2.5 Flash-Lite Preview (06-17), Arena rating: 1,187 (scale 750.0 to 1325.6), 95% confidence interval 1179.9 to 1194.71,187(1179.9–1194.7)
  102. 102
    StepFunDossier

    Arena rating

    Step-1o Turbo (202506), Arena rating: 1,186 (scale 750.0 to 1325.6), 95% confidence interval 1171.4 to 1199.81,186(1171.4–1199.8)
  103. 103
    OpenAIDossier

    Arena rating

    GPT-4.1 mini, Arena rating: 1,181 (scale 750.0 to 1325.6), 95% confidence interval 1173.8 to 1188.71,181(1173.8–1188.7)
  104. 104
    TencentDossier

    Arena rating

    Hunyuan Large Vision, Arena rating: 1,180 (scale 750.0 to 1325.6), 95% confidence interval 1163.4 to 1196.11,180(1163.4–1196.1)
  105. 105
    TencentDossier

    Arena rating

    Hunyuan Vision 1.5 Thinking, Arena rating: 1,179 (scale 750.0 to 1325.6), 95% confidence interval 1166.8 to 1191.11,179(1166.8–1191.1)
  106. 106
    StepFunDossier

    Arena rating

    Step-3, Arena rating: 1,176 (scale 750.0 to 1325.6), 95% confidence interval 1164.3 to 1187.71,176(1164.3–1187.7)
  107. 107
    AnthropicDossier

    Arena rating

    Opus 4, Arena rating: 1,175 (scale 750.0 to 1325.6), 95% confidence interval 1162.3 to 1188.21,175(1162.3–1188.2)
  108. 108
    AnthropicDossier

    Arena rating

    Claude Sonnet 4, Arena rating: 1,175 (scale 750.0 to 1325.6), 95% confidence interval 1161.1 to 1189.21,175(1161.1–1189.2)
  109. 109
    MistralDossier

    Arena rating

    Mistral Medium 3.1, Arena rating: 1,172 (scale 750.0 to 1325.6), 95% confidence interval 1166.2 to 1178.51,172(1166.2–1178.5)
  110. 110
    AnthropicDossier

    thinking · 32,000 budget

    Arena rating

    Claude 3.7 Sonnet, Arena rating: 1,168 (scale 750.0 to 1325.6), 95% confidence interval 1152.9 to 1184.01,168(1152.9–1184.0)
  111. 111
    OpenAIDossier

    Arena rating

    o1, Arena rating: 1,168 (scale 750.0 to 1325.6), 95% confidence interval 1157.5 to 1179.01,168(1157.5–1179.0)
  112. 112
    GoogleDossier

    Arena rating

    Gemma 3 27B, Arena rating: 1,164 (scale 750.0 to 1325.6), 95% confidence interval 1156.3 to 1172.41,164(1156.3–1172.4)
  113. 113

    Arena rating

    GLM-4.6V, Arena rating: 1,163 (scale 750.0 to 1325.6), 95% confidence interval 1149.4 to 1176.81,163(1149.4–1176.8)
  114. 114
    GoogleDossier

    Arena rating

    Gemini 1.5 Pro (002), Arena rating: 1,161 (scale 750.0 to 1325.6), 95% confidence interval 1152.2 to 1170.41,161(1152.2–1170.4)
  115. 115
    OpenAIDossier

    high

    Arena rating

    GPT-5 nano, Arena rating: 1,159 (scale 750.0 to 1325.6), 95% confidence interval 1148.0 to 1170.11,159(1148.0–1170.1)
  116. 116
    GoogleDossier

    Arena rating

    Gemini 2.0 Flash (001), Arena rating: 1,158 (scale 750.0 to 1325.6), 95% confidence interval 1150.5 to 1164.91,158(1150.5–1164.9)
  117. 117
    MistralDossier

    Arena rating

    Mistral Medium 3, Arena rating: 1,157 (scale 750.0 to 1325.6), 95% confidence interval 1148.4 to 1165.01,157(1148.4–1165.0)
  118. 118

    Arena rating

    GLM-4.5V, Arena rating: 1,154 (scale 750.0 to 1325.6), 95% confidence interval 1142.5 to 1165.71,154(1142.5–1165.7)
  119. 119
    AlibabaDossier

    Arena rating

    Qwen2.5 VL 32B Instruct, Arena rating: 1,152 (scale 750.0 to 1325.6), 95% confidence interval 1136.6 to 1167.41,152(1136.6–1167.4)
  120. 120
    AnthropicDossier

    Arena rating

    Claude 3.7 Sonnet, Arena rating: 1,151 (scale 750.0 to 1325.6), 95% confidence interval 1141.4 to 1160.11,151(1141.4–1160.1)
  121. 121
    MistralDossier

    Arena rating

    Mistral Small 3.2, Arena rating: 1,142 (scale 750.0 to 1325.6), 95% confidence interval 1132.5 to 1150.71,142(1132.5–1150.7)
  122. 122

    Arena rating

    Llama 4 Maverick, Arena rating: 1,142 (scale 750.0 to 1325.6), 95% confidence interval 1132.5 to 1150.61,142(1132.5–1150.6)
  123. 123
    GoogleDossier

    Arena rating

    Gemini 1.5 Flash (002), Arena rating: 1,141 (scale 750.0 to 1325.6), 95% confidence interval 1130.6 to 1150.51,141(1130.6–1150.5)
  124. 124
    OpenAIDossier

    Arena rating

    GPT-4o (May '24), Arena rating: 1,137 (scale 750.0 to 1325.6), 95% confidence interval 1127.7 to 1145.31,137(1127.7–1145.3)
  125. 125
    MistralDossier

    Arena rating

    Mistral Small 3.1, Arena rating: 1,136 (scale 750.0 to 1325.6), 95% confidence interval 1127.6 to 1144.61,136(1127.6–1144.6)
  126. 126
    AnthropicDossier

    Arena rating

    Claude 3.5 Sonnet (Oct '24), Arena rating: 1,125 (scale 750.0 to 1325.6), 95% confidence interval 1117.5 to 1133.41,125(1117.5–1133.4)
  127. 127
    StepFunDossier

    high-res

    Arena rating

    Step-1o Vision 32K, Arena rating: 1,122 (scale 750.0 to 1325.6), 95% confidence interval 1109.0 to 1134.41,122(1109.0–1134.4)
  128. 128
    AnthropicDossier

    Arena rating

    Claude 3.5 Sonnet (June '24), Arena rating: 1,120 (scale 750.0 to 1325.6), 95% confidence interval 1110.1 to 1130.31,120(1110.1–1130.3)
  129. 129

    Arena rating

    Llama 4 Scout, Arena rating: 1,118 (scale 750.0 to 1325.6), 95% confidence interval 1108.9 to 1127.81,118(1108.9–1127.8)
  130. 130
    AlibabaDossier

    Arena rating

    Qwen2.5 VL 72B Instruct, Arena rating: 1,107 (scale 750.0 to 1325.6), 95% confidence interval 1096.4 to 1118.21,107(1096.4–1118.2)
  131. 131
    GoogleDossier

    Arena rating

    Gemini 2.0 Flash-Lite Preview (02-05), Arena rating: 1,100 (scale 750.0 to 1325.6), 95% confidence interval 1089.8 to 1110.81,100(1089.8–1110.8)
  132. 132
    AnthropicDossier

    Arena rating

    Claude 3.5 Haiku, Arena rating: 1,092 (scale 750.0 to 1325.6), 95% confidence interval 1075.9 to 1108.01,092(1075.9–1108.0)
  133. 133
    OpenAIDossier

    Arena rating

    GPT-4 Turbo, Arena rating: 1,090 (scale 750.0 to 1325.6), 95% confidence interval 1078.2 to 1101.41,090(1078.2–1101.4)
  134. 134
    MistralDossier

    Arena rating

    Pixtral Large (2411), Arena rating: 1,089 (scale 750.0 to 1325.6), 95% confidence interval 1079.9 to 1098.61,089(1079.9–1098.6)
  135. 135
    GoogleDossier

    Arena rating

    Gemini 1.5 Pro (001), Arena rating: 1,088 (scale 750.0 to 1325.6), 95% confidence interval 1076.5 to 1099.91,088(1076.5–1099.9)
  136. 136
    Allen Institute for AIDossier

    Arena rating

    Molmo 2 8B, Arena rating: 1,082 (scale 750.0 to 1325.6), 95% confidence interval 1062.0 to 1101.21,082(1062.0–1101.2)
  137. 137
    OpenAIDossier

    Arena rating

    GPT-4o mini, Arena rating: 1,066 (scale 750.0 to 1325.6), 95% confidence interval 1057.2 to 1073.91,066(1057.2–1073.9)
  138. 138
    OpenAIDossier

    Arena rating

    GPT-4o (Aug '24), Arena rating: 1,064 (scale 750.0 to 1325.6), 95% confidence interval 1051.9 to 1077.01,064(1051.9–1077.0)
  139. 139
    OpenAIDossier

    Arena rating

    GPT-4.1 nano, Arena rating: 1,063 (scale 750.0 to 1325.6), 95% confidence interval 1044.8 to 1081.41,063(1044.8–1081.4)
  140. 140
    AlibabaDossier

    Arena rating

    Qwen VL Max (1119), Arena rating: 1,057 (scale 750.0 to 1325.6), 95% confidence interval 1041.2 to 1073.71,057(1041.2–1073.7)
  141. 141
    AlibabaDossier

    Arena rating

    Qwen2 VL 72B, Arena rating: 1,047 (scale 750.0 to 1325.6), 95% confidence interval 1036.6 to 1057.11,047(1036.6–1057.1)
  142. 142
    StepFunDossier

    Arena rating

    Step-1V 32K, Arena rating: 1,046 (scale 750.0 to 1325.6), 95% confidence interval 1029.2 to 1062.61,046(1029.2–1062.6)
  143. 143
    GoogleDossier

    Arena rating

    Gemini 1.5 Flash-8B (001), Arena rating: 1,044 (scale 750.0 to 1325.6), 95% confidence interval 1033.3 to 1054.61,044(1033.3–1054.6)
  144. 144
    AnthropicDossier

    Arena rating

    Claude 3 Opus, Arena rating: 1,023 (scale 750.0 to 1325.6), 95% confidence interval 1012.4 to 1033.21,023(1012.4–1033.2)
  145. 145
    Allen Institute for AIDossier

    Arena rating

    Molmo 72B (0924), Arena rating: 1,012 (scale 750.0 to 1325.6), 95% confidence interval 999.1 to 1025.31,012(999.1–1025.3)
  146. 146
    MistralDossier

    Arena rating

    Pixtral 12B (2409), Arena rating: 1,008 (scale 750.0 to 1325.6), 95% confidence interval 998.9 to 1018.01,008(998.9–1018.0)
  147. 147
    GoogleDossier

    Arena rating

    Gemini 1.5 Flash (001), Arena rating: 1,008 (scale 750.0 to 1325.6), 95% confidence interval 996.0 to 1019.71,008(996.0–1019.7)
  148. 148
    Shanghai AI LaboratoryDossier

    Arena rating

    InternVL2 26B, Arena rating: 1,004 (scale 750.0 to 1325.6), 95% confidence interval 991.6 to 1017.31,004(991.6–1017.3)
  149. 149

    Arena rating

    Llama 3.2 90B Vision Instruct, Arena rating: 1,000 (scale 750.0 to 1325.6), 95% confidence interval 990.3 to 1008.81,000(990.3–1008.8)
  150. 150
    TencentDossier

    Arena rating

    Hunyuan Standard Vision, Arena rating: 999 (scale 750.0 to 1325.6), 95% confidence interval 977.6 to 1020.4999(977.6–1020.4)
  151. 151
    CohereDossier

    Arena rating

    Aya Vision 32B, Arena rating: 995 (scale 750.0 to 1325.6), 95% confidence interval 973.5 to 1017.5995(973.5–1017.5)
  152. 152
    AmazonDossier

    Arena rating

    Nova Lite, Arena rating: 990 (scale 750.0 to 1325.6), 95% confidence interval 975.3 to 1005.3990(975.3–1005.3)
  153. 153
    AlibabaDossier

    Arena rating

    Qwen2 VL 7B Instruct, Arena rating: 990 (scale 750.0 to 1325.6), 95% confidence interval 979.2 to 1000.4990(979.2–1000.4)
  154. 154
    AnthropicDossier

    Arena rating

    Claude 3 Sonnet, Arena rating: 984 (scale 750.0 to 1325.6), 95% confidence interval 972.4 to 995.2984(972.4–995.2)
  155. 155
    AmazonDossier

    Arena rating

    Nova Pro, Arena rating: 980 (scale 750.0 to 1325.6), 95% confidence interval 966.4 to 994.2980(966.4–994.2)
  156. 156
    01.AIDossier

    Arena rating

    Yi Vision, Arena rating: 978 (scale 750.0 to 1325.6), 95% confidence interval 959.5 to 995.8978(959.5–995.8)
  157. 157

    Arena rating

    Llama 3.2 Instruct 11B (Vision), Arena rating: 966 (scale 750.0 to 1325.6), 95% confidence interval 954.9 to 977.6966(954.9–977.6)
  158. 158
    Allen Institute for AIDossier

    Arena rating

    Molmo 7B-D (0924), Arena rating: 959 (scale 750.0 to 1325.6), 95% confidence interval 944.8 to 972.4959(944.8–972.4)
  159. 159
    AnthropicDossier

    Arena rating

    Claude 3 Haiku, Arena rating: 950 (scale 750.0 to 1325.6), 95% confidence interval 937.5 to 962.6950(937.5–962.6)
  160. 160
    Shanghai AI LaboratoryDossier

    Arena rating

    InternVL2 4B, Arena rating: 941 (scale 750.0 to 1325.6), 95% confidence interval 928.7 to 953.8941(928.7–953.8)
  161. 161
    NVIDIADossier

    Arena rating

    NVILA Internal 15B v1, Arena rating: 940 (scale 750.0 to 1325.6), 95% confidence interval 920.2 to 959.9940(920.2–959.9)
  162. 162
    LLaVA TeamDossier

    Arena rating

    LLaVA 1.6 34B, Arena rating: 936 (scale 750.0 to 1325.6), 95% confidence interval 924.1 to 948.6936(924.1–948.6)
  163. 163

    Arena rating

    CogVLM2 Llama3 Chat 19B, Arena rating: 927 (scale 750.0 to 1325.6), 95% confidence interval 912.2 to 942.7927(912.2–942.7)
  164. 164
    LLaVA TeamDossier

    Arena rating

    LLaVA-OneVision Qwen2 72B OV, Arena rating: 923 (scale 750.0 to 1325.6), 95% confidence interval 904.7 to 940.7923(904.7–940.7)
  165. 165
    OpenBMBDossier

    Arena rating

    MiniCPM-V 2.6, Arena rating: 912 (scale 750.0 to 1325.6), 95% confidence interval 895.5 to 927.5912(895.5–927.5)
  166. 166
    MicrosoftDossier

    Arena rating

    Phi-3.5 Vision Instruct, Arena rating: 851 (scale 750.0 to 1325.6), 95% confidence interval 835.6 to 867.2851(835.6–867.2)
  167. 167
    MicrosoftDossier

    Arena rating

    Phi-3 Vision 128K Instruct, Arena rating: 812 (scale 750.0 to 1325.6), 95% confidence interval 792.9 to 830.3812(792.9–830.3)