Ir para o conteúdo
AI.info

Model Tracker

Epoch Capabilities Index leaderboard

The models in the Model Tracker that Epoch AI has measured on Epoch Capabilities Index, ranked by the highest score of each: the name Epoch AI gives it, the setting it ran and the day.

Em inglês

The models in the Model Tracker that Epoch AI has measured on Epoch Capabilities Index, ranked by the highest score of each: the name Epoch AI gives it, the setting it ran and the day. Epoch AI, ‘Capabilities & benchmarking’, epoch.ai/benchmarks, CC BY 4.0: Epoch AI’s own runs and its Capabilities Index.

  1. 1. Claude Fable 5.2: 173.5 (170.1–177.4) — Claude Fable 5.2, 20 November 2026
  2. 2. GPT-6 Astra: 169.2 (166.3–172.5) — GPT-6 Astra, 20 November 2026
  3. 3. Claude Opus 5.5: 168.7 (165.9–171.9) — Claude Opus 5.5, 20 November 2026
  4. 4. GPT-6.1 Sol: 168.6 (165.7–171.9) — GPT-6.1 Sol, 20 November 2026
  5. 5. Claude Fable 5.1: 167.1 (164.4–170.1) — Claude Fable 5.1, 20 November 2026
  6. 6. Claude Sonnet 5.5: 166.6 (163.8–169.7) — Claude Sonnet 5.5, 20 November 2026
  7. 7. Gemini 4 Argon: 166.6 (163.9–169.5) — Gemini 4 Argon, 20 November 2026
  8. 8. Claude Opus 5: 164.7 (162.3–167.7) — Claude Opus 5, 20 November 2026
  9. 9. Claude Fable 5: 164.2 (161.8–166.9) — Claude Fable 5, 20 November 2026
  10. 10. GPT-6 Sol: 163.8 (161.4–166.6) — GPT-6 Sol, 20 November 2026
  11. 11. GPT-5.6 Sol: 163.1 (160.8–165.6) — GPT-5.6 Sol, 20 November 2026
  12. 12. GPT-5.5 Pro: 162.8 (160.1–165.8) — GPT-5.5 Pro, 20 November 2026
  13. 13. Muse Wonder: 161.7 (159.4–164.3) — Muse Wonder, 20 November 2026
  14. 14. GPT-5.6 Terra: 160.8 (158.6–163.2) — GPT-5.6 Terra, 20 November 2026
  15. 15. GPT-5.4 Pro: 160.3 (157.9–163.0) — GPT-5.4 Pro, 20 November 2026
  16. 16. Claude Haiku 5.5: 160.3 (158.1–162.7) — Claude Haiku 5.5, 20 November 2026
  17. 17. GPT-5.5: 160.2 (158.2–162.4) — GPT-5.5, 20 November 2026
  18. 18. Claude Opus 4.8: 159.9 (157.8–162.2) — Claude Opus 4.8, 20 November 2026
  19. 19. Gemini 3.8 Flash: 158.8 (156.9–161.0) — Gemini 3.8 Flash, 20 November 2026
  20. 20. Gemini 3.7 Flash: 158.4 (156.4–160.5) — Gemini 3.7 Flash, 20 November 2026
  21. 21. GPT-5.4: 157.9 (156.0–160.0) — GPT-5.4, 20 November 2026
  22. 22. Grok 4.6: 157.8 (155.9–160.0) — Grok 4.6, 20 November 2026
  23. 23. Qwen3.8-Max: 157.6 (155.5–159.9) — Qwen 3.8 Max, 20 November 2026
  24. 24. Claude Opus 4.7: 157.5 (155.6–159.5) — Claude Opus 4.7, 20 November 2026
  25. 25. GPT-5.2 Pro: 157.3 (155.1–159.8) — GPT-5.2 Pro, 20 November 2026
  26. 26. DeepSeek-V4.1-Flash: 157.0 (155.1–159.0) — DeepSeek V4.1 Flash, 20 November 2026
  27. 27. Claude Sonnet 5: 156.9 (154.9–159.1) — Claude Sonnet 5, 20 November 2026
  28. 28. GPT-5.6 Luna: 156.7 (154.9–158.8) — GPT-5.6 Luna, 20 November 2026
  29. 29. Grok 4.7: 156.4 (154.5–158.5) — Grok 4.7, 20 November 2026
  30. 30. GPT-6 Luna: 156.4 (154.5–158.5) — GPT-6 Luna, 20 November 2026
  31. 31. Claude Opus 4.6: 156.4 (154.6–158.3) — Claude Opus 4.6, 20 November 2026
  32. 32. Gemini 3.1 Pro: 156.3 (154.5–158.2) — Gemini 3.1 Pro, 20 November 2026
  33. 33. Gemini 3.5 Flash: 155.4 (153.5–157.4) — Gemini 3.5 Flash, 20 November 2026
  34. 34. Gemini 3.6 Flash: 155.2 (153.4–157.2) — Gemini 3.6 Flash, 20 November 2026
  35. 35. Grok 4.5: 155.0 (153.2–156.9) — Grok 4.5, 20 November 2026
  36. 36. Mistral Large 4: 154.6 (151.8–157.4) — Mistral Large 4, 20 November 2026
  37. 37. GPT-5.2: 154.5 (152.9–156.3) — GPT-5.2, 20 November 2026
  38. 38. Muse Spark: 153.2 (150.9–155.5) — Muse Spark, 20 November 2026
  39. 39. Gemini 3 Pro: 153.0 (151.5–154.8) — Gemini 3 Pro, 20 November 2026
  40. 40. Claude Sonnet 4.6: 152.9 (151.2–154.7) — Claude Sonnet 4.6, 20 November 2026
  41. 41. Qwen3.8-27B: 152.7 (150.7–154.8) — Qwen 3.8 27B, 20 November 2026
  42. 42. Gemini 3 Flash: 152.3 (150.8–154.0) — Gemini 3 Flash, 20 November 2026
  43. 43. Grok 4.20: 152.3 (150.6–154.1) — Grok 4.20, 20 November 2026
  44. 44. Claude Opus 4.5: 151.9 (150.3–153.5) — Claude Opus 4.5, 20 November 2026
  45. 45. GPT-5 Pro: 151.3 (149.3–153.3) — GPT-5 Pro, 20 November 2026
  46. 46. GPT-5.1: 151.0 (149.4–152.6) — GPT-5.1, 20 November 2026
  47. 47. DeepSeek V4 Pro: 150.2 (148.5–152.0) — DeepSeek-V4-Pro, 20 November 2026
  48. 48. GPT-5: 150.0 — GPT-5, 20 November 2026
  49. 49. GPT-5.4 Mini: 149.4 (147.7–151.1) — GPT-5.4 Mini, 20 November 2026
  50. 50. Claude Sonnet 4.5: 147.6 (146.1–149.0) — Claude Sonnet 4.5, 20 November 2026
  51. 51. o3-pro: 147.5 (145.6–149.4) — o3-pro, 20 November 2026
  52. 52. o3: 147.1 (145.6–148.6) — o3, 20 November 2026
  53. 53. Grok 4: 147.0 (145.5–148.5) — Grok 4, 20 November 2026
  54. 54. DeepSeek-V4-Flash: 146.8 (143.6–150.2) — DeepSeek-V4-Flash, 20 November 2026
  55. 55. Qwen3.6-27B: 146.8 (144.7–148.9) — Qwen3.6 27B, 20 November 2026
  56. 56. GPT-5.4 nano: 146.3 (144.7–148.0) — GPT-5.4 Nano, 20 November 2026
  57. 57. GPT-5 Mini: 146.0 (144.4–147.6) — GPT-5 mini, 20 November 2026
  58. 58. Gemini 3.5 Flash-Lite: 145.7 (144.0–147.3) — Gemini 3.5 Flash-Lite, 20 November 2026
  59. 59. o4-mini: 145.2 (143.7–146.7) — o4-mini, 20 November 2026
  60. 60. Gemini 3.1 Flash-Lite: 145.0 (142.9–147.3) — Gemini 3.1 Flash-Lite, 20 November 2026
  61. 61. Gemini 2.5 Pro: 144.9 (143.5–146.3) — Gemini 2.5 Pro (Jun 2025), 20 November 2026
  62. 62. Claude Opus 4.1: 144.8 (143.0–146.5) — Claude Opus 4.1, 20 November 2026
  63. 63. DeepSeek-V3.2-Exp: 144.4 (142.1–146.7) — DeepSeek-V3.2-Exp, 20 November 2026
  64. 64. Grok 4 Fast: 144.4 (142.2–146.5) — Grok 4 Fast, 20 November 2026
  65. 65. Qwen3.6-35B-A3B: 144.1 (142.2–146.0) — Qwen 3.6 35B-A3B, 20 November 2026
  66. 66. Qwen3.5-35B-A3B: 144.1 (141.6–146.7) — Qwen3.5-35B-A3B, 20 November 2026
  67. 67. Claude Haiku 4.5: 142.8 (141.2–144.3) — Claude Haiku 4.5, 20 November 2026
  68. 68. Gemini 2.5 Flash: 141.9 (139.1–144.6) — Gemini 2.5 Flash (Sep 2025), 20 November 2026
  69. 69. Claude Sonnet 4: 141.9 (140.3–143.4) — Claude Sonnet 4, 20 November 2026
  70. 70. Mistral Medium 3.5: 141.6 (138.4–144.7) — Mistral Medium 3.5, 20 November 2026
  71. 71. o1: 141.2 (139.5–142.8) — o1, 20 November 2026
  72. 72. Claude Sonnet 3.7: 141.1 (139.4–142.6) — Claude 3.7 Sonnet, 20 November 2026
  73. 73. DeepSeek-V3.1: 140.8 (137.6–143.9) — DeepSeek-V3.1, 20 November 2026
  74. 74. DeepSeek-R1: 140.8 (138.9–142.5) — DeepSeek-R1 (May 2025), 20 November 2026
  75. 75. o3-mini: 140.7 (139.0–142.3) — o3-mini, 20 November 2026
  76. 76. gpt-oss-120b: 140.2 (138.3–142.1) — gpt-oss-120b, 20 November 2026
  77. 77. Grok 3 mini: 140.0 (138.2–141.7) — Grok-3 mini, 20 November 2026
  78. 78. Qwen3.5-9B: 139.5 (137.3–141.7) — Qwen3.5-9B, 20 November 2026
  79. 79. GPT-5 nano: 139.1 (137.2–140.9) — GPT-5 nano, 20 November 2026
  80. 80. Qwen3-235B-A22B: 138.4 (136.1–140.7) — Qwen3-235B-A22B, 20 November 2026
  81. 81. Grok 3: 137.6 (135.6–139.4) — Grok 3, 20 November 2026
  82. 82. Qwen3-32B: 136.8 (134.4–139.2) — Qwen3-32B, 20 November 2026
  83. 83. GPT-4.5: 136.5 (134.4–138.5) — GPT-4.5, 20 November 2026
  84. 84. GPT-4.1: 136.2 (134.1–138.1) — GPT-4.1, 20 November 2026
  85. 85. Qwen3-14B: 136.1 (133.4–138.6) — Qwen3-14B, 20 November 2026
  86. 86. o1-mini: 135.3 (133.1–137.3) — o1-mini, 20 November 2026
  87. 86. DeepSeek-V3: 135.3 (133.1–137.3) — DeepSeek-V3 (Mar 2025), 20 November 2026
  88. 88. gpt-oss-20b: 134.9 (132.3–137.3) — gpt-oss-20b, 20 November 2026
  89. 89. GPT-4.1 mini: 134.3 (132.1–136.2) — GPT-4.1 mini, 20 November 2026
  90. 90. Qwen3-30B-A3B: 134.2 (131.5–136.8) — Qwen3-30B-A3B, 20 November 2026
  91. 91. Gemini 2.0 Flash: 134.0 (131.7–136.2) — Gemini 2.0 Flash (Feb 2025), 20 November 2026
  92. 92. o1-preview: 133.5 (131.0–135.9) — o1-preview, 20 November 2026
  93. 93. Gemini 2.5 Flash-Lite: 133.1 (129.4–136.5) — Gemini 2.5 Flash-Lite (Jun 2025), 20 November 2026
  94. 94. Mistral Medium 3: 132.8 (130.2–135.2) — Mistral Medium 3, 20 November 2026
  95. 95. Llama 4 Maverick: 132.2 (130.0–134.3) — Llama 4 Maverick, 20 November 2026
  96. 96. Qwen3-8B: 132.1 (129.1–134.8) — Qwen3-8B, 20 November 2026
  97. 97. Gemini 1.5 Pro: 130.5 (127.8–132.8) — Gemini 1.5 Pro (Sept 2024), 20 November 2026
  98. 98. Grok-2: 129.8 (127.0–132.3) — Grok-2 (Dec 2024), 20 November 2026
  99. 99. GPT-4o: 128.8 (126.1–131.2) — GPT-4o (Nov 2024), 20 November 2026
  100. 100. Llama 4 Scout: 127.8 (124.9–130.4) — Llama 4 Scout, 20 November 2026
  101. 101. Gemini 1.5 Flash: 127.0 (123.9–129.8) — Gemini 1.5 Flash (Sep 2024), 20 November 2026
  102. 102. GPT-4.1 nano: 126.8 (123.8–129.6) — GPT-4.1 nano, 20 November 2026
  103. 103. GPT-4 Turbo: 125.8 (122.8–128.6) — GPT-4 Turbo (Apr 2024), 20 November 2026
  104. 104. Claude 3 Opus: 125.6 (122.7–128.2) — Claude 3 Opus, 20 November 2026
  105. 105. Claude 3.5 Haiku: 124.8 (121.5–127.7) — Claude 3.5 Haiku, 20 November 2026
  106. 106. GPT-4o mini: 123.8 (120.5–126.7) — GPT-4o mini, 20 November 2026
  107. 107. GPT-4: 121.6 (117.7–125.0) — GPT-4 (Mar 2023), 20 November 2026
  108. 108. Claude 3 Sonnet: 119.7 (115.7–123.2) — Claude 3 Sonnet, 20 November 2026
  109. 109. Mixtral 8x22B: 118.7 (114.3–122.5) — Mixtral 8x22B, 20 November 2026
  110. 110. Claude 3 Haiku: 117.8 (113.6–121.6) — Claude 3 Haiku, 20 November 2026
  111. 111. Claude 2: 117.1 (112.6–121.1) — Claude 2, 20 November 2026
  112. 112. Mixtral 8x7B: 113.1 (108.1–117.5) — Mixtral 8x7B, 20 November 2026
  113. 113. GPT-3.5 Turbo: 112.8 (107.8–117.2) — GPT-3.5 Turbo (Jan 2024), 20 November 2026
  114. 114. Qwen-14B: 101.6 (92.8–109.5) — Qwen-14B, 20 November 2026
  115. 115. Qwen-7B: 88.9 (77.5–98.6) — Qwen-7B, 20 November 2026

Epoch AI: the benchmark and its data · CC BY 4.0

Other benchmarks

All of the Model Tracker