Benchmark Almanac

AGI-Eval Community Instruction-Tuned Language Models

GPT-5.5 0.86 leads since April 2026
Not in model cards

Models answer Chinese and English questions across reasoning, knowledge, learning, interaction and instruction following.

The 2026/11 leaderboard reports 142 scored model entries on its primary measure.

Released
March 2025
Kind
Reasoning in many languages
Level
School
Scored
Checked automatically
Answers
Multiple choice
Tasks
Kept private
Measures
Capability
Language
Chinese, English
In and out
Text → Text
Interaction
Single turn
Refresh
One fixed set
Run by
A host
Strongest model tested
Claude Fable 5 #12 on the capability index
Last model added
July 2026
Scores refreshed
November 21st, 2026
122
Entries on its own leaderboard, 94 models we track

GPT-5.5 leads with 0.86 among models released by July 2026 GPT-5.5 leads with 0.86 among models released by July 2026

Best score per entry from the benchmark's own leaderboard, read on November 21st, 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. This view shows only models we track, each at its best.

The newest model with a score here came out in July 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

122 entries. Scroll sideways for the rest.

  1. 0.86 GPT-5.5
  2. 0.85 Claude Fable 5
  3. 0.84 Gemini 3.1 Pro
  4. 0.84 Kimi K3
  5. 0.83 Gemini 3.5 Flash
  6. 0.83 Doubao Seed 2.1 Pro
  7. 0.82 Claude Opus 4.8
  8. 0.82 GPT-5.4
  9. 0.82 Qwen3.7-Max
  10. 0.81 Seed2.0 Pro
  11. 0.81 DeepSeek-V4-Pro
  12. 0.81 Kimi K2.6
  13. 0.81 Claude Opus 4.6
  14. 0.80 Claude Sonnet 5
  15. 0.80 GLM-5.2
  16. 0.80 DeepSeek-V4-Flash
  17. 0.80 Claude Opus 4.5
  18. 0.79 GPT-5.1
  19. 0.79 Gemini 3 Flash
  20. 0.79 Gemini 3 Pro
  21. 0.79 GPT-5-Thinking (High)
  22. 0.78 Tencent Hy3
  23. 0.78 Qwen3.5 397B-A17B
  24. 0.77 o3-pro
  25. 0.77 GLM-5.1
  26. 0.77 Kimi K2.5
  27. 0.77 DeepSeek-V3.2-Thinking-Speciale
  28. 0.76 Tencent Hy3 preview
  29. 0.76 Qwen3-Max
  30. 0.75 Qwen 3.6 Plus
  31. 0.74 Doubao-Seed-1.8
  32. 0.74 o3
  33. 0.74 GLM-5
  34. 0.74 gpt-oss-120b
  35. 0.73 Doubao-Seed-1.6-Thinking-2507
  36. 0.73 Qwen3-Max-Preview
  37. 0.73 DeepSeek-V3.2
  38. 0.73 MiniMax-M3
  39. 0.72 Kimi K2 Thinking
  40. 0.72 Gemini 2.5 Pro
  41. 0.72 GLM-4.7
  42. 0.71 Qwen3-235B-A22B-Thinking
  43. 0.71 Grok 4
  44. 0.71 o4-mini
  45. 0.71 o3-mini
  46. 0.70 gpt-oss-20b
  47. 0.70 LongCat-2.0
  48. 0.70 Grok 4.1
  49. 0.70 ERNIE 5.0
  50. 0.69 DeepSeek-R1
  51. 0.69 DeepSeek-V3.1
  52. 0.69 Doubao-Seed-1.6
  53. 0.69 Ernie-5.0-Thinking-Preview
  54. 0.69 DeepSeek-V3.2-Exp
  55. 0.69 Claude Sonnet 4.5
  56. 0.69 DeepSeek-V3.1-Terminus
  57. 0.68 Doubao-Seed-1.6-Auto
  58. 0.68 Claude-Opus-4.5-no-reasoning
  59. 0.68 o1
  60. 0.68 Doubao-Seed-1.6-On
  61. 0.68 LongCat-Flash-Thinking-2601
  62. 0.67 MiniMax-M2
  63. 0.67 LongCat-Flash-Thinking-2509
  64. 0.67 Qwen3-Next-80B-A3B
  65. 0.67 Gemini 2.5 Flash
  66. 0.67 Doubao-1.5-Thinking-Pro-2504
  67. 0.67 Kimi-K2.5-Chat
  68. 0.66 Qwen3-Max-Instruct
  69. 0.65 Qwen3-Max-no-Thinking
  70. 0.65 GPT-5.2
  71. 0.65 GLM-4.5
  72. 0.65 GLM-4.6
  73. 0.64 Qwen3-235B-A22B
  74. 0.64 Ring-1T
  75. 0.64 Claude Opus 4
  76. 0.64 LongCat-Flash-2512
  77. 0.63 Qwen3-235B-A22B-Instruct
  78. 0.62 LongCat-Flash-2510
  79. 0.62 GPT-5.1 Instant
  80. 0.62 Kimi K2
  81. 0.62 Qwen3-32B
  82. 0.62 Step 3
  83. 0.62 Qwen3-Next-80B-A3B-Instruct
  84. 0.61 Doubao-Seed-1.6-Off
  85. 0.61 LongCat-Flash-2508
  86. 0.60 GPT-5
  87. 0.60 Claude Sonnet 4
  88. 0.60 Qwen3-30B-A3B-Thinking
  89. 0.59 QwQ-32B
  90. 0.59 DeepSeek-V3
  91. 0.59 MiniMax-M1-80k
  92. 0.59 Qwen3-14B
  93. 0.58 Hunyuan-T1
  94. 0.58 GPT-4.5
  95. 0.55 Doubao-Seed-1.6-Flash
  96. 0.55 Claude 3.7 Sonnet
  97. 0.55 GPT-4.1
  98. 0.55 Step-r1-v-mini
  99. 0.55 Qwen3-8B
  100. 0.55 Doubao-1.5-Pro-32k
  101. 0.53 ERNIE-4.5-300B-A47B
  102. 0.52 Grok 3
  103. 0.50 Llama 4 Maverick
  104. 0.48 Qwen3-30B-A3B
  105. 0.48 Qwen3-4B
  106. 0.48 Claude-3.5-Sonnet-1022
  107. 0.46 GPT-4o-1120
  108. 0.46 GLM4-Plus
  109. 0.45 Qwen2.5-72B
  110. 0.44 Qwen2.5-32B-Instruct
  111. 0.43 Gemma 3 27B
  112. 0.41 Llama 4 Scout
  113. 0.41 Llama3.1-405B-Instruct
  114. 0.40 Qwen2.5-14B-Instruct
  115. 0.38 Qwen3-1.7B
  116. 0.37 Gemma 3 12B
  117. 0.35 abab6.5
  118. 0.34 Qwen2.5-7B-Instruct
  119. 0.31 GLM-4-9B
  120. 0.27 Gemma 3 4B
  121. 0.23 Qwen3-0.6B
  122. 0.15 Gemma 3 1B
  1. 1 GPT-5.5 0.86
  2. 2 Claude Fable 5 0.85
  3. 3 Gemini 3.1 Pro 0.84
  4. 4 Kimi K3 0.84
  5. 5 Gemini 3.5 Flash 0.83
  6. 6 Doubao Seed 2.1 Pro 0.83
  7. 7 Claude Opus 4.8 0.82
  8. 8 GPT-5.4 0.82
  9. 9 Qwen3.7-Max 0.82
  10. 10 Seed2.0 Pro 0.81
  11. 11 DeepSeek-V4-Pro max 0.81
  12. 12 Kimi K2.6 0.81
  13. 13 Claude Opus 4.6 0.81
  14. 14 Claude Sonnet 5 0.80
  15. 15 GLM-5.2 0.80
  16. 16 DeepSeek-V4-Flash max 0.80
  17. 17 Claude Opus 4.5 0.80
  18. 18 GPT-5.1 0.79
  19. 19 Gemini 3 Flash 0.79
  20. 20 Gemini 3 Pro 0.79
  21. 21 GPT-5-Thinking (High) 0.79
  22. 22 Tencent Hy3 0.78
  23. 23 Qwen3.5 397B-A17B thinking 0.78
  24. 24 o3-pro 0.77
  25. 25 GLM-5.1 0.77
  26. 26 Kimi K2.5 thinking 0.77
  27. 27 DeepSeek-V3.2-Thinking-Speciale 0.77
  28. 28 Tencent Hy3 preview 0.76
  29. 29 Qwen3-Max 0.76
  30. 30 Qwen 3.6 Plus 0.75
  31. 31 Doubao-Seed-1.8 0.74
  32. 32 o3 0.74
  33. 33 GLM-5 0.74
  34. 34 gpt-oss-120b high 0.74
  35. 35 Doubao-Seed-1.6-Thinking-2507 0.73
  36. 36 Qwen3-Max-Preview thinking 0.73
  37. 37 DeepSeek-V3.2 0.73
  38. 38 MiniMax-M3 0.73
  39. 39 Kimi K2 Thinking 0.72
  40. 40 Gemini 2.5 Pro Jun 2025 0.72
  41. 41 GLM-4.7 0.72
  42. 42 Qwen3-235B-A22B-Thinking Jul 2025 0.71
  43. 43 Grok 4 0.71
  44. 44 o4-mini 0.71
  45. 45 o3-mini high 0.71
  46. 46 gpt-oss-20b high 0.70
  47. 47 LongCat-2.0 0.70
  48. 48 Grok 4.1 0.70
  49. 49 ERNIE 5.0 0.70
  50. 50 DeepSeek-R1 May 2025 0.69
  51. 51 DeepSeek-V3.1 0.69
  52. 52 Doubao-Seed-1.6 thinking 0.69
  53. 53 Ernie-5.0-Thinking-Preview 0.69
  54. 54 DeepSeek-V3.2-Exp thinking 0.69
  55. 55 Claude Sonnet 4.5 thinking 0.69
  56. 56 DeepSeek-V3.1-Terminus thinking 0.69
  57. 57 Doubao-Seed-1.6-Auto 0.68
  58. 58 Claude-Opus-4.5-no-reasoning 0.68
  59. 59 o1 0.68
  60. 60 Doubao-Seed-1.6-On 0.68
  61. 61 LongCat-Flash-Thinking-2601 0.68
  62. 62 MiniMax-M2 0.67
  63. 63 LongCat-Flash-Thinking-2509 0.67
  64. 64 Qwen3-Next-80B-A3B thinking 0.67
  65. 65 Gemini 2.5 Flash Sep 2025 0.67
  66. 66 Doubao-1.5-Thinking-Pro-2504 0.67
  67. 67 Kimi-K2.5-Chat 0.67
  68. 68 Qwen3-Max-Instruct 0.66
  69. 69 Qwen3-Max-no-Thinking 0.65
  70. 70 GPT-5.2 0.65
  71. 71 GLM-4.5 0.65
  72. 72 GLM-4.6 0.65
  73. 73 Qwen3-235B-A22B thinking 0.64
  74. 74 Ring-1T 0.64
  75. 75 Claude Opus 4 thinking 0.64
  76. 76 LongCat-Flash-2512 0.64
  77. 77 Qwen3-235B-A22B-Instruct Jul 2025 0.63
  78. 78 LongCat-Flash-2510 0.62
  79. 79 GPT-5.1 Instant 0.62
  80. 80 Kimi K2 Jul 2025 0.62
  81. 81 Qwen3-32B thinking 0.62
  82. 82 Step 3 0.62
  83. 83 Qwen3-Next-80B-A3B-Instruct 0.62
  84. 84 Doubao-Seed-1.6-Off 0.61
  85. 85 LongCat-Flash-2508 0.61
  86. 86 GPT-5 0.60
  87. 87 Claude Sonnet 4 thinking 0.60
  88. 88 Qwen3-30B-A3B-Thinking Jul 2025 0.60
  89. 89 QwQ-32B 0.59
  90. 90 DeepSeek-V3 Mar 2025 0.59
  91. 91 MiniMax-M1-80k 0.59
  92. 92 Qwen3-14B thinking 0.59
  93. 93 Hunyuan-T1 0.58
  94. 94 GPT-4.5 0.58
  95. 95 Doubao-Seed-1.6-Flash 0.55
  96. 96 Claude 3.7 Sonnet thinking 0.55
  97. 97 GPT-4.1 0.55
  98. 98 Step-r1-v-mini 0.55
  99. 99 Qwen3-8B thinking 0.55
  100. 100 Doubao-1.5-Pro-32k 0.55
  101. 101 ERNIE-4.5-300B-A47B 0.53
  102. 102 Grok 3 0.52
  103. 103 Llama 4 Maverick 0.50
  104. 104 Qwen3-30B-A3B 0.48
  105. 105 Qwen3-4B thinking 0.48
  106. 106 Claude-3.5-Sonnet-1022 0.48
  107. 107 GPT-4o-1120 0.46
  108. 108 GLM4-Plus 0.46
  109. 109 Qwen2.5-72B 0.45
  110. 110 Qwen2.5-32B-Instruct 0.44
  111. 111 Gemma 3 27B 0.43
  112. 112 Llama 4 Scout 0.41
  113. 113 Llama3.1-405B-Instruct 0.41
  114. 114 Qwen2.5-14B-Instruct 0.40
  115. 115 Qwen3-1.7B thinking 0.38
  116. 116 Gemma 3 12B 0.37
  117. 117 abab6.5 0.35
  118. 118 Qwen2.5-7B-Instruct 0.34
  119. 119 GLM-4-9B 0.31
  120. 120 Gemma 3 4B 0.27
  121. 121 Qwen3-0.6B thinking 0.23
  122. 122 Gemma 3 1B 0.15

Source: Official AGI-Eval Community Instruction-Tuned Language Models leaderboard

94 models. Scroll sideways for the rest.

  1. 0.86 GPT-5.5
  2. 0.85 Claude Fable 5
  3. 0.84 Gemini 3.1 Pro
  4. 0.84 Kimi K3
  5. 0.83 Gemini 3.5 Flash
  6. 0.83 Doubao Seed 2.1 Pro
  7. 0.82 Claude Opus 4.8
  8. 0.82 GPT-5.4
  9. 0.82 Qwen3.7-Max
  10. 0.81 Seed2.0 Pro
  11. 0.81 DeepSeek-V4-Pro
  12. 0.81 Kimi K2.6
  13. 0.81 Claude Opus 4.6
  14. 0.80 Claude Sonnet 5
  15. 0.80 GLM-5.2
  16. 0.80 DeepSeek-V4-Flash
  17. 0.80 Claude Opus 4.5
  18. 0.79 GPT-5.1
  19. 0.79 Gemini 3 Flash
  20. 0.79 Gemini 3 Pro
  21. 0.78 Tencent Hy3
  22. 0.78 Qwen3.5 397B-A17B
  23. 0.77 o3-pro
  24. 0.77 GLM-5.1
  25. 0.77 Kimi K2.5
  26. 0.76 Tencent Hy3 preview
  27. 0.76 Qwen3-Max
  28. 0.75 Qwen 3.6 Plus
  29. 0.74 o3
  30. 0.74 GLM-5
  31. 0.74 gpt-oss-120b
  32. 0.73 Qwen3-Max-Preview
  33. 0.73 DeepSeek-V3.2
  34. 0.73 MiniMax-M3
  35. 0.72 Kimi K2 Thinking
  36. 0.72 Gemini 2.5 Pro
  37. 0.72 GLM-4.7
  38. 0.71 Qwen3-235B-A22B-Thinking
  39. 0.71 Grok 4
  40. 0.71 o4-mini
  41. 0.71 o3-mini
  42. 0.70 gpt-oss-20b
  43. 0.70 LongCat-2.0
  44. 0.70 Grok 4.1
  45. 0.70 ERNIE 5.0
  46. 0.69 DeepSeek-R1
  47. 0.69 DeepSeek-V3.1
  48. 0.69 Doubao-Seed-1.6
  49. 0.69 DeepSeek-V3.2-Exp
  50. 0.69 Claude Sonnet 4.5
  51. 0.69 DeepSeek-V3.1-Terminus
  52. 0.68 o1
  53. 0.68 LongCat-Flash-Thinking-2601
  54. 0.67 MiniMax-M2
  55. 0.67 Qwen3-Next-80B-A3B
  56. 0.67 Gemini 2.5 Flash
  57. 0.65 GPT-5.2
  58. 0.65 GLM-4.5
  59. 0.65 GLM-4.6
  60. 0.64 Qwen3-235B-A22B
  61. 0.64 Ring-1T
  62. 0.64 Claude Opus 4
  63. 0.63 Qwen3-235B-A22B-Instruct
  64. 0.62 GPT-5.1 Instant
  65. 0.62 Kimi K2
  66. 0.62 Qwen3-32B
  67. 0.62 Step 3
  68. 0.62 Qwen3-Next-80B-A3B-Instruct
  69. 0.60 GPT-5
  70. 0.60 Claude Sonnet 4
  71. 0.60 Qwen3-30B-A3B-Thinking
  72. 0.59 QwQ-32B
  73. 0.59 DeepSeek-V3
  74. 0.59 MiniMax-M1-80k
  75. 0.59 Qwen3-14B
  76. 0.58 Hunyuan-T1
  77. 0.58 GPT-4.5
  78. 0.55 Claude 3.7 Sonnet
  79. 0.55 GPT-4.1
  80. 0.55 Qwen3-8B
  81. 0.53 ERNIE-4.5-300B-A47B
  82. 0.52 Grok 3
  83. 0.50 Llama 4 Maverick
  84. 0.48 Qwen3-30B-A3B
  85. 0.48 Qwen3-4B
  86. 0.45 Qwen2.5-72B
  87. 0.43 Gemma 3 27B
  88. 0.41 Llama 4 Scout
  89. 0.38 Qwen3-1.7B
  90. 0.37 Gemma 3 12B
  91. 0.31 GLM-4-9B
  92. 0.27 Gemma 3 4B
  93. 0.23 Qwen3-0.6B
  94. 0.15 Gemma 3 1B
  1. 1 GPT-5.5 0.86
  2. 2 Claude Fable 5 0.85
  3. 3 Gemini 3.1 Pro 0.84
  4. 4 Kimi K3 0.84
  5. 5 Gemini 3.5 Flash 0.83
  6. 6 Doubao Seed 2.1 Pro 0.83
  7. 7 Claude Opus 4.8 0.82
  8. 8 GPT-5.4 0.82
  9. 9 Qwen3.7-Max 0.82
  10. 10 Seed2.0 Pro 0.81
  11. 11 DeepSeek-V4-Pro max 0.81
  12. 12 Kimi K2.6 0.81
  13. 13 Claude Opus 4.6 0.81
  14. 14 Claude Sonnet 5 0.80
  15. 15 GLM-5.2 0.80
  16. 16 DeepSeek-V4-Flash max 0.80
  17. 17 Claude Opus 4.5 0.80
  18. 18 GPT-5.1 0.79
  19. 19 Gemini 3 Flash 0.79
  20. 20 Gemini 3 Pro 0.79
  21. 21 Tencent Hy3 0.78
  22. 22 Qwen3.5 397B-A17B thinking 0.78
  23. 23 o3-pro 0.77
  24. 24 GLM-5.1 0.77
  25. 25 Kimi K2.5 thinking 0.77
  26. 26 Tencent Hy3 preview 0.76
  27. 27 Qwen3-Max 0.76
  28. 28 Qwen 3.6 Plus 0.75
  29. 29 o3 0.74
  30. 30 GLM-5 0.74
  31. 31 gpt-oss-120b high 0.74
  32. 32 Qwen3-Max-Preview thinking 0.73
  33. 33 DeepSeek-V3.2 0.73
  34. 34 MiniMax-M3 0.73
  35. 35 Kimi K2 Thinking 0.72
  36. 36 Gemini 2.5 Pro Jun 2025 0.72
  37. 37 GLM-4.7 0.72
  38. 38 Qwen3-235B-A22B-Thinking Jul 2025 0.71
  39. 39 Grok 4 0.71
  40. 40 o4-mini 0.71
  41. 41 o3-mini high 0.71
  42. 42 gpt-oss-20b high 0.70
  43. 43 LongCat-2.0 0.70
  44. 44 Grok 4.1 0.70
  45. 45 ERNIE 5.0 0.70
  46. 46 DeepSeek-R1 May 2025 0.69
  47. 47 DeepSeek-V3.1 0.69
  48. 48 Doubao-Seed-1.6 thinking 0.69
  49. 49 DeepSeek-V3.2-Exp thinking 0.69
  50. 50 Claude Sonnet 4.5 thinking 0.69
  51. 51 DeepSeek-V3.1-Terminus thinking 0.69
  52. 52 o1 0.68
  53. 53 LongCat-Flash-Thinking-2601 0.68
  54. 54 MiniMax-M2 0.67
  55. 55 Qwen3-Next-80B-A3B thinking 0.67
  56. 56 Gemini 2.5 Flash Sep 2025 0.67
  57. 57 GPT-5.2 0.65
  58. 58 GLM-4.5 0.65
  59. 59 GLM-4.6 0.65
  60. 60 Qwen3-235B-A22B thinking 0.64
  61. 61 Ring-1T 0.64
  62. 62 Claude Opus 4 thinking 0.64
  63. 63 Qwen3-235B-A22B-Instruct Jul 2025 0.63
  64. 64 GPT-5.1 Instant 0.62
  65. 65 Kimi K2 Jul 2025 0.62
  66. 66 Qwen3-32B thinking 0.62
  67. 67 Step 3 0.62
  68. 68 Qwen3-Next-80B-A3B-Instruct 0.62
  69. 69 GPT-5 0.60
  70. 70 Claude Sonnet 4 thinking 0.60
  71. 71 Qwen3-30B-A3B-Thinking Jul 2025 0.60
  72. 72 QwQ-32B 0.59
  73. 73 DeepSeek-V3 Mar 2025 0.59
  74. 74 MiniMax-M1-80k 0.59
  75. 75 Qwen3-14B thinking 0.59
  76. 76 Hunyuan-T1 0.58
  77. 77 GPT-4.5 0.58
  78. 78 Claude 3.7 Sonnet thinking 0.55
  79. 79 GPT-4.1 0.55
  80. 80 Qwen3-8B thinking 0.55
  81. 81 ERNIE-4.5-300B-A47B 0.53
  82. 82 Grok 3 0.52
  83. 83 Llama 4 Maverick 0.50
  84. 84 Qwen3-30B-A3B 0.48
  85. 85 Qwen3-4B thinking 0.48
  86. 86 Qwen2.5-72B 0.45
  87. 87 Gemma 3 27B 0.43
  88. 88 Llama 4 Scout 0.41
  89. 89 Qwen3-1.7B thinking 0.38
  90. 90 Gemma 3 12B 0.37
  91. 91 GLM-4-9B 0.31
  92. 92 Gemma 3 4B 0.27
  93. 93 Qwen3-0.6B thinking 0.23
  94. 94 Gemma 3 1B 0.15

Source: Official AGI-Eval Community Instruction-Tuned Language Models leaderboard

#EntryScoreSettingReleasedRun onSource
1 GPT-5.5 0.86 Apr 2026 May 2026 Official leaderboard
2 Claude Fable 5 0.85 Jun 2026 Jul 2026 Official leaderboard
3 Gemini 3.1 Pro 0.84 Feb 2026 Mar 2026 Official leaderboard
4 Kimi K3 0.84 Jul 2026 Jul 2026 Official leaderboard
5 Gemini 3.5 Flash 0.83 May 2026 May 2026 Official leaderboard
6 Doubao Seed 2.1 Pro 0.83 Jun 2026 Jul 2026 Official leaderboard
7 Claude Opus 4.8 0.82 May 2026 Jun 2026 Official leaderboard
8 GPT-5.4 0.82 Mar 2026 Apr 2026 Official leaderboard
9 Qwen3.7-Max 0.82 May 2026 May 2026 Official leaderboard
10 Seed2.0 Pro 0.81 Feb 2026 Mar 2026 Official leaderboard
11 DeepSeek-V4-Pro 0.81 max Apr 2026 May 2026 Official leaderboard
12 Kimi K2.6 0.81 Apr 2026 May 2026 Official leaderboard
13 Claude Opus 4.6 0.81 Feb 2026 Mar 2026 Official leaderboard
14 Claude Sonnet 5 0.80 Jun 2026 Jul 2026 Official leaderboard
15 GLM-5.2 0.80 Jun 2026 Jul 2026 Official leaderboard
16 DeepSeek-V4-Flash 0.80 max Apr 2026 May 2026 Official leaderboard
17 Claude Opus 4.5 0.80 Nov 2025 Feb 2026 Official leaderboard
18 GPT-5.1 0.79 Nov 2025 Dec 2025 Official leaderboard
19 Gemini 3 Flash 0.79 Dec 2025 Dec 2025 Official leaderboard
20 Gemini 3 Pro 0.79 Nov 2025 Dec 2025 Official leaderboard
21 GPT-5-Thinking (High) 0.79 Feb 2026 Official leaderboard
22 Tencent Hy3 0.78 Jul 2026 Jul 2026 Official leaderboard
23 Qwen3.5 397B-A17B 0.78 thinking Feb 2026 Mar 2026 Official leaderboard
24 o3-pro 0.77 Jun 2025 Feb 2026 Official leaderboard
25 GLM-5.1 0.77 Apr 2026 Jun 2026 Official leaderboard
26 Kimi K2.5 0.77 thinking Jan 2026 Feb 2026 Official leaderboard
27 DeepSeek-V3.2-Thinking-Speciale 0.77 Mar 2026 Official leaderboard
28 Tencent Hy3 preview 0.76 Apr 2026 May 2026 Official leaderboard
29 Qwen3-Max 0.76 Sep 2025 Feb 2026 Official leaderboard
30 Qwen 3.6 Plus 0.75 Mar 2026 Jun 2026 Official leaderboard
31 Doubao-Seed-1.8 0.74 Jan 2026 Official leaderboard
32 o3 0.74 Apr 2025 Mar 2026 Official leaderboard
33 GLM-5 0.74 Feb 2026 Mar 2026 Official leaderboard
34 gpt-oss-120b 0.74 high Aug 2025 Mar 2026 Official leaderboard
35 Doubao-Seed-1.6-Thinking-2507 0.73 Mar 2026 Official leaderboard
36 Qwen3-Max-Preview 0.73 thinking Sep 2025 Nov 2025 Official leaderboard
37 DeepSeek-V3.2 0.73 Dec 2025 Dec 2025 Official leaderboard
38 MiniMax-M3 0.73 Jun 2026 Jun 2026 Official leaderboard
39 Kimi K2 Thinking 0.72 Nov 2025 Nov 2025 Official leaderboard
40 Gemini 2.5 Pro Jun 2025 0.72 Jun 2025 Mar 2026 Official leaderboard
41 GLM-4.7 0.72 Dec 2025 Dec 2025 Official leaderboard
42 Qwen3-235B-A22B-Thinking Jul 2025 0.71 Jul 2025 Mar 2026 Official leaderboard
43 Grok 4 0.71 Jul 2025 Mar 2026 Official leaderboard
44 o4-mini 0.71 Apr 2025 Mar 2026 Official leaderboard
45 o3-mini 0.71 high Jan 2025 Mar 2026 Official leaderboard
46 gpt-oss-20b 0.70 high Aug 2025 Mar 2026 Official leaderboard
47 LongCat-2.0 0.70 Jul 2026 Jul 2026 Official leaderboard
48 Grok 4.1 0.70 Nov 2025 Mar 2026 Official leaderboard
49 ERNIE 5.0 0.70 Jan 2026 Feb 2026 Official leaderboard
50 DeepSeek-R1 May 2025 0.69 May 2025 Mar 2026 Official leaderboard
51 DeepSeek-V3.1 0.69 Aug 2025 Mar 2026 Official leaderboard
52 Doubao-Seed-1.6 0.69 thinking Jun 2025 Mar 2026 Official leaderboard
53 Ernie-5.0-Thinking-Preview 0.69 Jan 2026 Official leaderboard
54 DeepSeek-V3.2-Exp 0.69 thinking Sep 2025 Mar 2026 Official leaderboard
55 Claude Sonnet 4.5 0.69 thinking Sep 2025 Mar 2026 Official leaderboard
56 DeepSeek-V3.1-Terminus 0.69 thinking Sep 2025 Mar 2026 Official leaderboard
57 Doubao-Seed-1.6-Auto 0.68 Mar 2026 Official leaderboard
58 Claude-Opus-4.5-no-reasoning 0.68 Feb 2026 Official leaderboard
59 o1 0.68 Dec 2024 Mar 2026 Official leaderboard
60 Doubao-Seed-1.6-On 0.68 Mar 2026 Official leaderboard
61 LongCat-Flash-Thinking-2601 0.68 Jan 2026 Feb 2026 Official leaderboard
62 MiniMax-M2 0.67 Oct 2025 Nov 2025 Official leaderboard
63 LongCat-Flash-Thinking-2509 0.67 Sep 2025 Official leaderboard
64 Qwen3-Next-80B-A3B 0.67 thinking Sep 2025 Mar 2026 Official leaderboard
65 Gemini 2.5 Flash Sep 2025 0.67 Sep 2025 Mar 2026 Official leaderboard
66 Doubao-1.5-Thinking-Pro-2504 0.67 Jan 2026 Official leaderboard
67 Kimi-K2.5-Chat 0.67 Feb 2026 Official leaderboard
68 Qwen3-Max-Instruct 0.66 Dec 2025 Official leaderboard
69 Qwen3-Max-no-Thinking 0.65 Feb 2026 Official leaderboard
70 GPT-5.2 0.65 Dec 2025 Mar 2026 Official leaderboard
71 GLM-4.5 0.65 Aug 2025 Mar 2026 Official leaderboard
72 GLM-4.6 0.65 Sep 2025 Mar 2026 Official leaderboard
73 Qwen3-235B-A22B 0.64 thinking Apr 2025 Mar 2026 Official leaderboard
74 Ring-1T 0.64 Oct 2025 Mar 2026 Official leaderboard
75 Claude Opus 4 0.64 thinking May 2025 Mar 2026 Official leaderboard
76 LongCat-Flash-2512 0.64 Dec 2025 Official leaderboard
77 Qwen3-235B-A22B-Instruct Jul 2025 0.63 Jul 2025 Mar 2026 Official leaderboard
78 LongCat-Flash-2510 0.62 Mar 2026 Official leaderboard
79 GPT-5.1 Instant 0.62 Nov 2025 Dec 2025 Official leaderboard
80 Kimi K2 Jul 2025 0.62 Jul 2025 Mar 2026 Official leaderboard
81 Qwen3-32B 0.62 thinking Apr 2025 Mar 2026 Official leaderboard
82 Step 3 0.62 Jul 2025 Mar 2026 Official leaderboard
83 Qwen3-Next-80B-A3B-Instruct 0.62 Sep 2025 Jan 2026 Official leaderboard
84 Doubao-Seed-1.6-Off 0.61 Feb 2026 Official leaderboard
85 LongCat-Flash-2508 0.61 Sep 2025 Official leaderboard
86 GPT-5 0.60 Aug 2025 Mar 2026 Official leaderboard
87 Claude Sonnet 4 0.60 thinking May 2025 Feb 2026 Official leaderboard
88 Qwen3-30B-A3B-Thinking Jul 2025 0.60 Jul 2025 Feb 2026 Official leaderboard
89 QwQ-32B 0.59 Mar 2025 Feb 2026 Official leaderboard
90 DeepSeek-V3 Mar 2025 0.59 Mar 2025 Feb 2026 Official leaderboard
91 MiniMax-M1-80k 0.59 Jun 2025 Sep 2025 Official leaderboard
92 Qwen3-14B 0.59 thinking Apr 2025 Mar 2026 Official leaderboard
93 Hunyuan-T1 0.58 Mar 2025 Mar 2026 Official leaderboard
94 GPT-4.5 0.58 Feb 2025 Jan 2026 Official leaderboard
95 Doubao-Seed-1.6-Flash 0.55 Feb 2026 Official leaderboard
96 Claude 3.7 Sonnet 0.55 thinking Feb 2025 Feb 2026 Official leaderboard
97 GPT-4.1 0.55 Apr 2025 Feb 2026 Official leaderboard
98 Step-r1-v-mini 0.55 Sep 2025 Official leaderboard
99 Qwen3-8B 0.55 thinking Apr 2025 Feb 2026 Official leaderboard
100 Doubao-1.5-Pro-32k 0.55 Mar 2026 Official leaderboard
101 ERNIE-4.5-300B-A47B 0.53 Jun 2025 Mar 2026 Official leaderboard
102 Grok 3 0.52 Apr 2025 Mar 2026 Official leaderboard
103 Llama 4 Maverick 0.50 Apr 2025 Mar 2026 Official leaderboard
104 Qwen3-30B-A3B 0.48 Apr 2025 Mar 2026 Official leaderboard
105 Qwen3-4B 0.48 thinking Apr 2025 Jan 2026 Official leaderboard
106 Claude-3.5-Sonnet-1022 0.48 Mar 2026 Official leaderboard
107 GPT-4o-1120 0.46 Mar 2026 Official leaderboard
108 GLM4-Plus 0.46 Mar 2026 Official leaderboard
109 Qwen2.5-72B 0.45 Sep 2024 Mar 2026 Official leaderboard
110 Qwen2.5-32B-Instruct 0.44 Mar 2026 Official leaderboard
111 Gemma 3 27B 0.43 Mar 2025 Mar 2026 Official leaderboard
112 Llama 4 Scout 0.41 Apr 2025 Mar 2026 Official leaderboard
113 Llama3.1-405B-Instruct 0.41 Mar 2026 Official leaderboard
114 Qwen2.5-14B-Instruct 0.40 Mar 2026 Official leaderboard
115 Qwen3-1.7B 0.38 thinking Apr 2025 Feb 2026 Official leaderboard
116 Gemma 3 12B 0.37 Mar 2025 Mar 2026 Official leaderboard
117 abab6.5 0.35 Jun 2026 Official leaderboard
118 Qwen2.5-7B-Instruct 0.34 Mar 2026 Official leaderboard
119 GLM-4-9B 0.31 Jun 2024 Mar 2026 Official leaderboard
120 Gemma 3 4B 0.27 Mar 2025 Feb 2026 Official leaderboard
121 Qwen3-0.6B 0.23 thinking Apr 2025 Mar 2026 Official leaderboard
122 Gemma 3 1B 0.15 Mar 2025 Mar 2026 Official leaderboard
#ModelScoreSettingReleasedRun onSource
1 GPT-5.5 0.86 Apr 2026 May 2026 Official leaderboard
2 Claude Fable 5 0.85 Jun 2026 Jul 2026 Official leaderboard
3 Gemini 3.1 Pro 0.84 Feb 2026 Mar 2026 Official leaderboard
4 Kimi K3 0.84 Jul 2026 Jul 2026 Official leaderboard
5 Gemini 3.5 Flash 0.83 May 2026 May 2026 Official leaderboard
6 Doubao Seed 2.1 Pro 0.83 Jun 2026 Jul 2026 Official leaderboard
7 Claude Opus 4.8 0.82 May 2026 Jun 2026 Official leaderboard
8 GPT-5.4 0.82 Mar 2026 Apr 2026 Official leaderboard
9 Qwen3.7-Max 0.82 May 2026 May 2026 Official leaderboard
10 Seed2.0 Pro 0.81 Feb 2026 Mar 2026 Official leaderboard
11 DeepSeek-V4-Pro 0.81 max Apr 2026 May 2026 Official leaderboard
12 Kimi K2.6 0.81 Apr 2026 May 2026 Official leaderboard
13 Claude Opus 4.6 0.81 Feb 2026 Mar 2026 Official leaderboard
14 Claude Sonnet 5 0.80 Jun 2026 Jul 2026 Official leaderboard
15 GLM-5.2 0.80 Jun 2026 Jul 2026 Official leaderboard
16 DeepSeek-V4-Flash 0.80 max Apr 2026 May 2026 Official leaderboard
17 Claude Opus 4.5 0.80 Nov 2025 Feb 2026 Official leaderboard
18 GPT-5.1 0.79 Nov 2025 Dec 2025 Official leaderboard
19 Gemini 3 Flash 0.79 Dec 2025 Dec 2025 Official leaderboard
20 Gemini 3 Pro 0.79 Nov 2025 Dec 2025 Official leaderboard
21 Tencent Hy3 0.78 Jul 2026 Jul 2026 Official leaderboard
22 Qwen3.5 397B-A17B 0.78 thinking Feb 2026 Mar 2026 Official leaderboard
23 o3-pro 0.77 Jun 2025 Feb 2026 Official leaderboard
24 GLM-5.1 0.77 Apr 2026 Jun 2026 Official leaderboard
25 Kimi K2.5 0.77 thinking Jan 2026 Feb 2026 Official leaderboard
26 Tencent Hy3 preview 0.76 Apr 2026 May 2026 Official leaderboard
27 Qwen3-Max 0.76 Sep 2025 Feb 2026 Official leaderboard
28 Qwen 3.6 Plus 0.75 Mar 2026 Jun 2026 Official leaderboard
29 o3 0.74 Apr 2025 Mar 2026 Official leaderboard
30 GLM-5 0.74 Feb 2026 Mar 2026 Official leaderboard
31 gpt-oss-120b 0.74 high Aug 2025 Mar 2026 Official leaderboard
32 Qwen3-Max-Preview 0.73 thinking Sep 2025 Nov 2025 Official leaderboard
33 DeepSeek-V3.2 0.73 Dec 2025 Dec 2025 Official leaderboard
34 MiniMax-M3 0.73 Jun 2026 Jun 2026 Official leaderboard
35 Kimi K2 Thinking 0.72 Nov 2025 Nov 2025 Official leaderboard
36 Gemini 2.5 Pro Jun 2025 0.72 Jun 2025 Mar 2026 Official leaderboard
37 GLM-4.7 0.72 Dec 2025 Dec 2025 Official leaderboard
38 Qwen3-235B-A22B-Thinking Jul 2025 0.71 Jul 2025 Mar 2026 Official leaderboard
39 Grok 4 0.71 Jul 2025 Mar 2026 Official leaderboard
40 o4-mini 0.71 Apr 2025 Mar 2026 Official leaderboard
41 o3-mini 0.71 high Jan 2025 Mar 2026 Official leaderboard
42 gpt-oss-20b 0.70 high Aug 2025 Mar 2026 Official leaderboard
43 LongCat-2.0 0.70 Jul 2026 Jul 2026 Official leaderboard
44 Grok 4.1 0.70 Nov 2025 Mar 2026 Official leaderboard
45 ERNIE 5.0 0.70 Jan 2026 Feb 2026 Official leaderboard
46 DeepSeek-R1 May 2025 0.69 May 2025 Mar 2026 Official leaderboard
47 DeepSeek-V3.1 0.69 Aug 2025 Mar 2026 Official leaderboard
48 Doubao-Seed-1.6 0.69 thinking Jun 2025 Mar 2026 Official leaderboard
49 DeepSeek-V3.2-Exp 0.69 thinking Sep 2025 Mar 2026 Official leaderboard
50 Claude Sonnet 4.5 0.69 thinking Sep 2025 Mar 2026 Official leaderboard
51 DeepSeek-V3.1-Terminus 0.69 thinking Sep 2025 Mar 2026 Official leaderboard
52 o1 0.68 Dec 2024 Mar 2026 Official leaderboard
53 LongCat-Flash-Thinking-2601 0.68 Jan 2026 Feb 2026 Official leaderboard
54 MiniMax-M2 0.67 Oct 2025 Nov 2025 Official leaderboard
55 Qwen3-Next-80B-A3B 0.67 thinking Sep 2025 Mar 2026 Official leaderboard
56 Gemini 2.5 Flash Sep 2025 0.67 Sep 2025 Mar 2026 Official leaderboard
57 GPT-5.2 0.65 Dec 2025 Mar 2026 Official leaderboard
58 GLM-4.5 0.65 Aug 2025 Mar 2026 Official leaderboard
59 GLM-4.6 0.65 Sep 2025 Mar 2026 Official leaderboard
60 Qwen3-235B-A22B 0.64 thinking Apr 2025 Mar 2026 Official leaderboard
61 Ring-1T 0.64 Oct 2025 Mar 2026 Official leaderboard
62 Claude Opus 4 0.64 thinking May 2025 Mar 2026 Official leaderboard
63 Qwen3-235B-A22B-Instruct Jul 2025 0.63 Jul 2025 Mar 2026 Official leaderboard
64 GPT-5.1 Instant 0.62 Nov 2025 Dec 2025 Official leaderboard
65 Kimi K2 Jul 2025 0.62 Jul 2025 Mar 2026 Official leaderboard
66 Qwen3-32B 0.62 thinking Apr 2025 Mar 2026 Official leaderboard
67 Step 3 0.62 Jul 2025 Mar 2026 Official leaderboard
68 Qwen3-Next-80B-A3B-Instruct 0.62 Sep 2025 Jan 2026 Official leaderboard
69 GPT-5 0.60 Aug 2025 Mar 2026 Official leaderboard
70 Claude Sonnet 4 0.60 thinking May 2025 Feb 2026 Official leaderboard
71 Qwen3-30B-A3B-Thinking Jul 2025 0.60 Jul 2025 Feb 2026 Official leaderboard
72 QwQ-32B 0.59 Mar 2025 Feb 2026 Official leaderboard
73 DeepSeek-V3 Mar 2025 0.59 Mar 2025 Feb 2026 Official leaderboard
74 MiniMax-M1-80k 0.59 Jun 2025 Sep 2025 Official leaderboard
75 Qwen3-14B 0.59 thinking Apr 2025 Mar 2026 Official leaderboard
76 Hunyuan-T1 0.58 Mar 2025 Mar 2026 Official leaderboard
77 GPT-4.5 0.58 Feb 2025 Jan 2026 Official leaderboard
78 Claude 3.7 Sonnet 0.55 thinking Feb 2025 Feb 2026 Official leaderboard
79 GPT-4.1 0.55 Apr 2025 Feb 2026 Official leaderboard
80 Qwen3-8B 0.55 thinking Apr 2025 Feb 2026 Official leaderboard
81 ERNIE-4.5-300B-A47B 0.53 Jun 2025 Mar 2026 Official leaderboard
82 Grok 3 0.52 Apr 2025 Mar 2026 Official leaderboard
83 Llama 4 Maverick 0.50 Apr 2025 Mar 2026 Official leaderboard
84 Qwen3-30B-A3B 0.48 Apr 2025 Mar 2026 Official leaderboard
85 Qwen3-4B 0.48 thinking Apr 2025 Jan 2026 Official leaderboard
86 Qwen2.5-72B 0.45 Sep 2024 Mar 2026 Official leaderboard
87 Gemma 3 27B 0.43 Mar 2025 Mar 2026 Official leaderboard
88 Llama 4 Scout 0.41 Apr 2025 Mar 2026 Official leaderboard
89 Qwen3-1.7B 0.38 thinking Apr 2025 Feb 2026 Official leaderboard
90 Gemma 3 12B 0.37 Mar 2025 Mar 2026 Official leaderboard
91 GLM-4-9B 0.31 Jun 2024 Mar 2026 Official leaderboard
92 Gemma 3 4B 0.27 Mar 2025 Feb 2026 Official leaderboard
93 Qwen3-0.6B 0.23 thinking Apr 2025 Mar 2026 Official leaderboard
94 Gemma 3 1B 0.15 Mar 2025 Mar 2026 Official leaderboard

Rankings by part

We read 15 parts from the benchmark’s own leaderboard in 2 groups: skill and language. Each ranks models on one part of the test. GPT-5.5 leads 5 of them, and 5 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.

Algorithm reasoning

94 models we track have a score here.

  1. 1 GPT-5.5 0.88
  2. 2 Claude Fable 5 0.86
  3. 3 Kimi K3 ↑1 0.81
  4. 4 Gemini 3.1 Pro ↓1 0.79
  5. 5 Claude Sonnet 5 ↑9 0.73
  6. 6 Doubao Seed 2.1 Pro 0.72
  7. 7 DeepSeek-V4-Pro ↑4 0.72
  8. 8 Gemini 3.5 Flash ↓3 0.71
  9. 9 GPT-5.4 ↓1 0.71
  10. 10 Gemini 3 Pro ↑10 0.70

Scores at each effort level

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
Qwen3-Max-Preview
thinking
0.73
DeepSeek-V3.2
non-thinking
0.73
o3-mini
medium high
0.71
DeepSeek-V3.2-Exp
thinking
0.69
Claude Sonnet 4.5
thinking
0.69
DeepSeek-V3.1-Terminus
thinking
0.69
Qwen3-235B-A22B
thinking
0.64
Claude Opus 4
thinking
0.64
Qwen3-32B
thinking
0.62
Claude Sonnet 4
thinking
0.60
Qwen3-14B
thinking
0.59
Claude 3.7 Sonnet
thinking
0.55

Source: Official AGI-Eval Community Instruction-Tuned Language Models leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
GPT-5.5 0.86
Claude Fable 5 0.85
Gemini 3.1 Pro 0.84
Kimi K3 0.84
Gemini 3.5 Flash 0.83
Doubao Seed 2.1 Pro 0.83
Claude Opus 4.8 0.82
GPT-5.4 0.82
Qwen3.7-Max 0.82
Seed2.0 Pro 0.81
DeepSeek-V4-Pro max 0.81
Kimi K2.6 0.81
Claude Opus 4.6 0.81
Claude Sonnet 5 0.80
GLM-5.2 0.80
DeepSeek-V4-Flash max 0.80
Claude Opus 4.5 0.80
GPT-5.1 0.79
Gemini 3 Flash 0.79
Gemini 3 Pro 0.79
Tencent Hy3 0.78
Qwen3.5 397B-A17B thinking 0.78
o3-pro 0.77
GLM-5.1 0.77
Kimi K2.5 thinking 0.77
Tencent Hy3 preview 0.76
Qwen3-Max 0.76
Qwen 3.6 Plus 0.75
o3 0.74
GLM-5 0.74
gpt-oss-120b high 0.74
Qwen3-Max-Preview thinking 0.73
DeepSeek-V3.2 0.73
MiniMax-M3 0.73
Kimi K2 Thinking 0.72
Gemini 2.5 Pro Jun 2025 0.72
GLM-4.7 0.72
Qwen3-235B-A22B-Thinking Jul 2025 0.71
Grok 4 0.71
o4-mini 0.71
o3-mini high 0.71
gpt-oss-20b high 0.70
LongCat-2.0 0.70
Grok 4.1 0.70
ERNIE 5.0 0.70
DeepSeek-R1 May 2025 0.69
DeepSeek-V3.1 0.69
Doubao-Seed-1.6 thinking 0.69
DeepSeek-V3.2-Exp thinking 0.69
Claude Sonnet 4.5 thinking 0.69
DeepSeek-V3.1-Terminus thinking 0.69
o1 0.68
LongCat-Flash-Thinking-2601 0.68
o3-mini medium 0.67
MiniMax-M2 0.67
Qwen3-Next-80B-A3B thinking 0.67
Gemini 2.5 Flash Sep 2025 0.67
Qwen3-Max-Preview 0.66
GPT-5.2 0.65
DeepSeek-R1 May 2025 0.65
GLM-4.5 0.65
GLM-4.6 0.65
Qwen3-235B-A22B-Thinking Jul 2025 0.65
Qwen3-235B-A22B thinking 0.64
Ring-1T 0.64
Claude Opus 4 thinking 0.64
DeepSeek-V3.2 non-thinking 0.63
DeepSeek-V3.2-Exp 0.63
DeepSeek-V3.1 0.63
Qwen3-235B-A22B-Instruct Jul 2025 0.63
DeepSeek-V3.1-Terminus 0.63
GPT-5.1 Instant 0.62
Kimi K2 Jul 2025 0.62
Qwen3-32B thinking 0.62
Step 3 0.62
Gemini 2.5 Flash Sep 2025 0.62
Qwen3-Next-80B-A3B-Instruct 0.62
Claude Sonnet 4.5 0.61
Claude Opus 4 0.61
GPT-5 0.60
Claude Sonnet 4 thinking 0.60
Qwen3-30B-A3B-Thinking Jul 2025 0.60
QwQ-32B 0.59
DeepSeek-V3 Mar 2025 0.59
MiniMax-M1-80k 0.59
Qwen3-14B thinking 0.59
Claude Sonnet 4 0.58
Hunyuan-T1 0.58
GPT-4.5 0.58
Claude 3.7 Sonnet thinking 0.55
GPT-4.1 0.55
Qwen3-8B thinking 0.55
ERNIE-4.5-300B-A47B 0.53
Qwen3-235B-A22B 0.53
Claude 3.7 Sonnet 0.52
Grok 3 0.52
Llama 4 Maverick 0.50
Qwen3-32B 0.49
Qwen3-30B-A3B 0.48
Qwen3-4B thinking 0.48
Qwen3-14B 0.47
Qwen2.5-72B 0.45
Qwen3-8B 0.43
Gemma 3 27B 0.43
Llama 4 Scout 0.41
Qwen3-4B 0.41
Qwen3-1.7B thinking 0.38
Gemma 3 12B 0.37
GLM-4-9B 0.31
Qwen3-1.7B 0.30
Gemma 3 4B 0.27
Qwen3-0.6B thinking 0.23
Qwen3-0.6B 0.19
Gemma 3 1B 0.15

Scores at each effort level

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
Qwen3-Max-Preview
thinking
0.73
DeepSeek-V3.2
non-thinking
0.73
o3-mini
medium high
0.71
DeepSeek-V3.2-Exp
thinking
0.69
Claude Sonnet 4.5
thinking
0.69
DeepSeek-V3.1-Terminus
thinking
0.69
Qwen3-235B-A22B
thinking
0.64
Claude Opus 4
thinking
0.64
Qwen3-32B
thinking
0.62
Claude Sonnet 4
thinking
0.60
Qwen3-14B
thinking
0.59
Claude 3.7 Sonnet
thinking
0.55

Source: Official AGI-Eval Community Instruction-Tuned Language Models leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
GPT-5.5 0.86
Claude Fable 5 0.85
Gemini 3.1 Pro 0.84
Kimi K3 0.84
Gemini 3.5 Flash 0.83
Doubao Seed 2.1 Pro 0.83
Claude Opus 4.8 0.82
GPT-5.4 0.82
Qwen3.7-Max 0.82
Seed2.0 Pro 0.81
DeepSeek-V4-Pro max 0.81
Kimi K2.6 0.81
Claude Opus 4.6 0.81
Claude Sonnet 5 0.80
GLM-5.2 0.80
DeepSeek-V4-Flash max 0.80
Claude Opus 4.5 0.80
GPT-5.1 0.79
Gemini 3 Flash 0.79
Gemini 3 Pro 0.79
Tencent Hy3 0.78
Qwen3.5 397B-A17B thinking 0.78
o3-pro 0.77
GLM-5.1 0.77
Kimi K2.5 thinking 0.77
Tencent Hy3 preview 0.76
Qwen3-Max 0.76
Qwen 3.6 Plus 0.75
o3 0.74
GLM-5 0.74
gpt-oss-120b high 0.74
Qwen3-Max-Preview thinking 0.73
DeepSeek-V3.2 0.73
MiniMax-M3 0.73
Kimi K2 Thinking 0.72
Gemini 2.5 Pro Jun 2025 0.72
GLM-4.7 0.72
Qwen3-235B-A22B-Thinking Jul 2025 0.71
Grok 4 0.71
o4-mini 0.71
o3-mini high 0.71
gpt-oss-20b high 0.70
LongCat-2.0 0.70
Grok 4.1 0.70
ERNIE 5.0 0.70
DeepSeek-R1 May 2025 0.69
DeepSeek-V3.1 0.69
Doubao-Seed-1.6 thinking 0.69
DeepSeek-V3.2-Exp thinking 0.69
Claude Sonnet 4.5 thinking 0.69
DeepSeek-V3.1-Terminus thinking 0.69
o1 0.68
LongCat-Flash-Thinking-2601 0.68
o3-mini medium 0.67
MiniMax-M2 0.67
Qwen3-Next-80B-A3B thinking 0.67
Gemini 2.5 Flash Sep 2025 0.67
Qwen3-Max-Preview 0.66
GPT-5.2 0.65
DeepSeek-R1 May 2025 0.65
GLM-4.5 0.65
GLM-4.6 0.65
Qwen3-235B-A22B-Thinking Jul 2025 0.65
Qwen3-235B-A22B thinking 0.64
Ring-1T 0.64
Claude Opus 4 thinking 0.64
DeepSeek-V3.2 non-thinking 0.63
DeepSeek-V3.2-Exp 0.63
DeepSeek-V3.1 0.63
Qwen3-235B-A22B-Instruct Jul 2025 0.63
DeepSeek-V3.1-Terminus 0.63
GPT-5.1 Instant 0.62
Kimi K2 Jul 2025 0.62
Qwen3-32B thinking 0.62
Step 3 0.62
Gemini 2.5 Flash Sep 2025 0.62
Qwen3-Next-80B-A3B-Instruct 0.62
Claude Sonnet 4.5 0.61
Claude Opus 4 0.61
GPT-5 0.60
Claude Sonnet 4 thinking 0.60
Qwen3-30B-A3B-Thinking Jul 2025 0.60
QwQ-32B 0.59
DeepSeek-V3 Mar 2025 0.59
MiniMax-M1-80k 0.59
Qwen3-14B thinking 0.59
Claude Sonnet 4 0.58
Hunyuan-T1 0.58
GPT-4.5 0.58
Claude 3.7 Sonnet thinking 0.55
GPT-4.1 0.55
Qwen3-8B thinking 0.55
ERNIE-4.5-300B-A47B 0.53
Qwen3-235B-A22B 0.53
Claude 3.7 Sonnet 0.52
Grok 3 0.52
Llama 4 Maverick 0.50
Qwen3-32B 0.49
Qwen3-30B-A3B 0.48
Qwen3-4B thinking 0.48
Qwen3-14B 0.47
Qwen2.5-72B 0.45
Qwen3-8B 0.43
Gemma 3 27B 0.43
Llama 4 Scout 0.41
Qwen3-4B 0.41
Qwen3-1.7B thinking 0.38
Gemma 3 12B 0.37
GLM-4-9B 0.31
Qwen3-1.7B 0.30
Gemma 3 4B 0.27
Qwen3-0.6B thinking 0.23
Qwen3-0.6B 0.19
Gemma 3 1B 0.15

9 models have held number one on the official leaderboard since its release in March 2025

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

OpenAI Anthropic ByteDance Seed Google DeepMind
Own leaderboard Not tested since Jul 2026
GPT-5.5 0.86

Source: Official AGI-Eval Community Instruction-Tuned Language Models leaderboard

Scores over time

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
0.00 0.20 0.40 0.60 0.80 1.00 2024 2025 2026 0.00 0.20 0.40 0.60 0.80 1.00 2024 2025 2026

Source: Official AGI-Eval Community Instruction-Tuned Language Models leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

More in Multilingual

All 545 in Multilingual
  • MMMLU In use MMLU professionally translated into 14 languages. Multiple-choice accuracy. Claude Mythos Preview 92.7%
  • WMT24++ In use English-to-55-language translation of WMT24 test data. Scored with automatic metrics such as MetricX or xCOMET. K-EXAONE 90.5%
  • MMLU-ProX In use MMLU-Pro translated into up to 29 languages with expert review. Multiple-choice accuracy per language. Qwen3.7-Max 87%
  • CLIcK In use Models answer 1,995 multiple-choice questions on Korean language and culture. Questions come from official Korean exams and textbooks in eleven categories. Solar Open 2 90.7%
  • Global MMLU Lite 3.0 In use Models answer 400 human-translated MMLU questions per language in 23 languages. Version 3.0 adds Czech, Hungarian, Oriya, Slovak and Tajik and updates Italian. Accuracy. Claude Opus 5 95.7%
  • MMMB In use Models answer multiple-choice questions about images in six languages across 15 categories. The set has about 12,000 questions. LFM2.5-VL-3B 83%

Data audit

All 9 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access The tasks are private.
  • Pass Confidence Checked against a primary source.
Scores 3 pass 2 notes
  • Pass Independent runs Scores from the official leaderboard (122 entries).
  • Pass Models scored 94 models we track, plus 28 other entries on its leaderboard.
  • Pass Newest score The newest is from July 2026, for Kimi K3.
  • Note Frontier models None of the top 10 has a score. The strongest tested is Claude Fable 5 (#12).
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on agi-eval.org: 114 rows for models we track and 28 other entries at the last read.
  • Pass Last read November 21st, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check