Benchmark Almanac

Linguistic Knowledge Benchmark (Fab AI)

GPT-5.4 97.8% leads since September 2026
Not in model cards

Models answer multiple-choice and short-answer questions about Kiswahili and English grammar, phonics, and spelling. Three model judges vote on short answers.

Fab AI scores 121 model configurations on 135 Kiswahili questions and 119 on 137 English questions. It shares 20 sample questions for each language.

Made by
Fab AI
Released
September 2026
Kind
Language reasoning
Level
University
Scored
Graded by a model
Answers
Multiple choice
Tasks
Public, plus a private test set
Measures
Capability
Language
Swahili, English
In and out
Text → Text
Interaction
Single turn
Items from
Exams and contests
Refresh
One fixed set
Run by
A host
Strongest model tested
Claude Opus 5 #11 on the capability index
Last model added
August 2026
Scores refreshed
November 20th, 2026
115
Entries on its own leaderboard, 109 models we track

GPT-5.4 leads with 97.8% among models released by August 2026 GPT-5.4 leads with 97.8% among models released by August 2026

Best score per entry from the benchmark's own leaderboard, read on November 20th, 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 August 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

115 entries. Scroll sideways for the rest.

  1. 97.8 GPT-5.4
  2. 97.0 Gemini 3.7 Flash
  3. 96.3 Gemini 3 Flash
  4. 95.6 Claude Opus 4.7 Adaptive Thinking
  5. 95.6 Gemini 3.1 Pro
  6. 95.6 Claude Opus 4.7
  7. 95.6 GPT-5.5
  8. 95.6 DeepSeek-V4-Pro
  9. 95.6 Gemini 3.5 Flash
  10. 95.6 Claude Fable 5
  11. 95.6 Grok 4.5
  12. 95.6 GPT-5.6 Sol
  13. 95.5 DeepSeek-V4-Flash
  14. 94.8 Claude Opus 4.8
  15. 94.1 Kimi K3
  16. 94.1 Grok 4.6
  17. 93.3 GPT-5.6 Terra
  18. 92.6 o3
  19. 92.6 Claude Opus 4.6
  20. 92.6 Grok 4.20
  21. 92.6 Inkling
  22. 92.6 Claude Opus 5
  23. 91.8 Gemini 2.5 Pro
  24. 91.8 GPT-5
  25. 91.8 Qwen 3.6 Max (Preview)
  26. 91.8 Muse Spark 1.2
  27. 91.1 Claude Opus 4.5
  28. 91.1 GPT-5.2
  29. 91.1 Gemini 3.1 Flash-Lite
  30. 91.1 GPT-5.4 Mini
  31. 91.1 Grok 4.3
  32. 91.1 Claude Sonnet 5
  33. 91.1 GPT-5.6 Luna
  34. 91.1 Gemini 3.6 Flash
  35. 90.4 Gemini 2.5 Flash
  36. 90.4 Claude Sonnet 4.6
  37. 90.4 Kimi K2.6
  38. 90.4 Gemini 3.5 Flash-Lite
  39. 89.6 GPT-5.1
  40. 89.6 Qwen3.7-Max
  41. 88.9 Inkling-Small
  42. 88.5 Qwen3.5 397B-A17B
  43. 88.2 Qwen3.8-2.4T-A95B
  44. 87.4 Kimi K2.5
  45. 87.4 Qwen 3.8 Max
  46. 86.7 Qwen3.5-Plus
  47. 86.7 MiMo-V2.5-Pro
  48. 85.9 Qwen 3.6 Plus
  49. 85.9 GLM-5.1
  50. 85.8 GLM-5.2
  51. 85.2 Gemma 4 31B IT
  52. 85.2 GLM-5.3
  53. 81.5 Claude Sonnet 4.5
  54. 80.7 GPT-4.1
  55. 80.7 GPT-5 mini
  56. 80.7 Qwen3.5-122B-A10B
  57. 80.0 sonar
  58. 80.0 Llama 4 Maverick
  59. 80.0 o4-mini
  60. 80.0 GPT-5.4 Nano
  61. 79.5 MiniMax-M3
  62. 79.3 Muse Glimmer
  63. 77.8 Gemma 4 26B A4B
  64. 77.8 MiMo-V2.5
  65. 76.3 Mistral Large 3
  66. 76.3 MiniMax-M2.7
  67. 74.8 Qwen3.5-27B
  68. 74.1 o3-mini
  69. 74.1 GPT-5 nano
  70. 74.1 Qwen3.5-Flash
  71. 73.9 Qwen3-235B-A22B-Thinking
  72. 73.1 Qwen 3.6 35B-A3B
  73. 72.6 DeepSeek-V3
  74. 71.8 gpt-oss-120b
  75. 71.6 Qwen3.5-35B-A3B
  76. 69.6 Qwen3.7 Flash
  77. 68.2 GPT-4.1 mini
  78. 67.9 Qwen 3.8 27B
  79. 66.9 Qwen3.6 27B
  80. 65.9 Claude Haiku 4.5
  81. 65.7 Gemma 4 12B
  82. 63.9 Doubao-Seed-1.6
  83. 63.7 Mistral Medium 3
  84. 62.2 Mistral Medium 3.1
  85. 62.2 Qwen 3.6 Flash
  86. 61.5 Mistral Small 4
  87. 60.0 Mistral Medium 3.5
  88. 59.3 GLM-4.6V
  89. 55.6 Qwen3-235B-A22B
  90. 54.8 Gemma 3 27B
  91. 54.1 Llama 4 Scout
  92. 54.1 GPT-4.1 nano
  93. 51.9 Qwen3.5-9B
  94. 50.4 Command A
  95. 46.6 glm-4.7-flash
  96. 45.9 Gemma 2 27B
  97. 42.2 Command-R+ August
  98. 40.7 Gemma 4 E2B
  99. 40.3 Nemotron 3.5 Lightning 30B-A3B
  100. 40.0 Mistral Small 3
  101. 39.3 Gemma 3 12B
  102. 38.9 Nemotron 3 Nano 30B-A3B
  103. 35.6 Nova 2 Lite
  104. 33.8 Seed 1.6 Flash
  105. 29.6 Mistral NeMo
  106. 28.9 Phi-4
  107. 26.7 Gemma 3 4B
  108. 23.7 Ministral 3 3B
  109. 23.0 Ministral 3 8B
  110. 22.2 Command R7B
  111. 21.5 Llama 3.2 3B
  112. 20.7 Qwen3-4B
  113. 14.1 Llama 3.2 1B
  114. 11.1 Qwen3-1.7B
  115. 10.4 Qwen3-0.6B
  1. 1 GPT-5.4 medium 97.8%
  2. 2 Gemini 3.7 Flash 97%
  3. 3 Gemini 3 Flash 96.3%
  4. 4 Claude Opus 4.7 Adaptive Thinking 95.6%
  5. 5 Gemini 3.1 Pro 95.6%
  6. 6 Claude Opus 4.7 95.6%
  7. 7 GPT-5.5 95.6%
  8. 8 DeepSeek-V4-Pro 95.6%
  9. 9 Gemini 3.5 Flash 95.6%
  10. 10 Claude Fable 5 95.6%
  11. 11 Grok 4.5 95.6%
  12. 12 GPT-5.6 Sol 95.6%
  13. 13 DeepSeek-V4-Flash 95.5%
  14. 14 Claude Opus 4.8 94.8%
  15. 15 Kimi K3 94.1%
  16. 16 Grok 4.6 94.1%
  17. 17 GPT-5.6 Terra 93.3%
  18. 18 o3 92.6%
  19. 19 Claude Opus 4.6 92.6%
  20. 20 Grok 4.20 reasoning 92.6%
  21. 21 Inkling 92.6%
  22. 22 Claude Opus 5 92.6%
  23. 23 Gemini 2.5 Pro Jun 2025 91.8%
  24. 24 GPT-5 medium 91.8%
  25. 25 Qwen 3.6 Max (Preview) 91.8%
  26. 26 Muse Spark 1.2 91.8%
  27. 27 Claude Opus 4.5 91.1%
  28. 28 GPT-5.2 medium 91.1%
  29. 29 Gemini 3.1 Flash-Lite 91.1%
  30. 30 GPT-5.4 Mini medium 91.1%
  31. 31 Grok 4.3 91.1%
  32. 32 Claude Sonnet 5 91.1%
  33. 33 GPT-5.6 Luna 91.1%
  34. 34 Gemini 3.6 Flash 91.1%
  35. 35 Gemini 2.5 Flash Sep 2025 90.4%
  36. 36 Claude Sonnet 4.6 90.4%
  37. 37 Kimi K2.6 90.4%
  38. 38 Gemini 3.5 Flash-Lite 90.4%
  39. 39 GPT-5.1 medium 89.6%
  40. 40 Qwen3.7-Max 89.6%
  41. 41 Inkling-Small 88.9%
  42. 42 Qwen3.5 397B-A17B 88.5%
  43. 43 Qwen3.8-2.4T-A95B 88.2%
  44. 44 Kimi K2.5 87.4%
  45. 45 Qwen 3.8 Max 87.4%
  46. 46 Qwen3.5-Plus 86.7%
  47. 47 MiMo-V2.5-Pro 86.7%
  48. 48 Qwen 3.6 Plus 85.9%
  49. 49 GLM-5.1 85.9%
  50. 50 GLM-5.2 85.8%
  51. 51 Gemma 4 31B IT 85.2%
  52. 52 GLM-5.3 85.2%
  53. 53 Claude Sonnet 4.5 81.5%
  54. 54 GPT-4.1 80.7%
  55. 55 GPT-5 mini medium 80.7%
  56. 56 Qwen3.5-122B-A10B 80.7%
  57. 57 sonar 80%
  58. 58 Llama 4 Maverick 80%
  59. 59 o4-mini 80%
  60. 60 GPT-5.4 Nano medium 80%
  61. 61 MiniMax-M3 79.5%
  62. 62 Muse Glimmer 79.3%
  63. 63 Gemma 4 26B A4B 77.8%
  64. 64 MiMo-V2.5 77.8%
  65. 65 Mistral Large 3 76.3%
  66. 66 MiniMax-M2.7 76.3%
  67. 67 Qwen3.5-27B 74.8%
  68. 68 o3-mini 74.1%
  69. 69 GPT-5 nano medium 74.1%
  70. 70 Qwen3.5-Flash 74.1%
  71. 71 Qwen3-235B-A22B-Thinking Jul 2025 73.9%
  72. 72 Qwen 3.6 35B-A3B 73.1%
  73. 73 DeepSeek-V3 Mar 2025 72.6%
  74. 74 gpt-oss-120b 71.8%
  75. 75 Qwen3.5-35B-A3B 71.6%
  76. 76 Qwen3.7 Flash 69.6%
  77. 77 GPT-4.1 mini 68.2%
  78. 78 Qwen 3.8 27B 67.9%
  79. 79 Qwen3.6 27B 66.9%
  80. 80 Claude Haiku 4.5 65.9%
  81. 81 Gemma 4 12B 65.7%
  82. 82 Doubao-Seed-1.6 63.9%
  83. 83 Mistral Medium 3 63.7%
  84. 84 Mistral Medium 3.1 62.2%
  85. 85 Qwen 3.6 Flash 62.2%
  86. 86 Mistral Small 4 61.5%
  87. 87 Mistral Medium 3.5 60%
  88. 88 GLM-4.6V 59.3%
  89. 89 Qwen3-235B-A22B 55.6%
  90. 90 Gemma 3 27B 54.8%
  91. 91 Llama 4 Scout 54.1%
  92. 92 GPT-4.1 nano 54.1%
  93. 93 Qwen3.5-9B 51.9%
  94. 94 Command A 50.4%
  95. 95 glm-4.7-flash 46.6%
  96. 96 Gemma 2 27B 45.9%
  97. 97 Command-R+ August 42.2%
  98. 98 Gemma 4 E2B 40.7%
  99. 99 Nemotron 3.5 Lightning 30B-A3B 40.3%
  100. 100 Mistral Small 3 40%
  101. 101 Gemma 3 12B 39.3%
  102. 102 Nemotron 3 Nano 30B-A3B 38.9%
  103. 103 Nova 2 Lite 35.6%
  104. 104 Seed 1.6 Flash 33.8%
  105. 105 Mistral NeMo 29.6%
  106. 106 Phi-4 28.9%
  107. 107 Gemma 3 4B 26.7%
  108. 108 Ministral 3 3B 23.7%
  109. 109 Ministral 3 8B 23%
  110. 110 Command R7B 22.2%
  111. 111 Llama 3.2 3B 21.5%
  112. 112 Qwen3-4B 20.7%
  113. 113 Llama 3.2 1B 14.1%
  114. 114 Qwen3-1.7B 11.1%
  115. 115 Qwen3-0.6B 10.4%

Source: Official Linguistic Knowledge Benchmark (Fab AI) leaderboard

109 models. Scroll sideways for the rest.

  1. 97.8 GPT-5.4
  2. 97.0 Gemini 3.7 Flash
  3. 96.3 Gemini 3 Flash
  4. 95.6 Gemini 3.1 Pro
  5. 95.6 Claude Opus 4.7
  6. 95.6 GPT-5.5
  7. 95.6 DeepSeek-V4-Pro
  8. 95.6 Gemini 3.5 Flash
  9. 95.6 Claude Fable 5
  10. 95.6 Grok 4.5
  11. 95.6 GPT-5.6 Sol
  12. 95.5 DeepSeek-V4-Flash
  13. 94.8 Claude Opus 4.8
  14. 94.1 Kimi K3
  15. 94.1 Grok 4.6
  16. 93.3 GPT-5.6 Terra
  17. 92.6 o3
  18. 92.6 Claude Opus 4.6
  19. 92.6 Grok 4.20
  20. 92.6 Inkling
  21. 92.6 Claude Opus 5
  22. 91.8 Gemini 2.5 Pro
  23. 91.8 GPT-5
  24. 91.8 Qwen 3.6 Max (Preview)
  25. 91.8 Muse Spark 1.2
  26. 91.1 Claude Opus 4.5
  27. 91.1 GPT-5.2
  28. 91.1 Gemini 3.1 Flash-Lite
  29. 91.1 GPT-5.4 Mini
  30. 91.1 Grok 4.3
  31. 91.1 Claude Sonnet 5
  32. 91.1 GPT-5.6 Luna
  33. 91.1 Gemini 3.6 Flash
  34. 90.4 Gemini 2.5 Flash
  35. 90.4 Claude Sonnet 4.6
  36. 90.4 Kimi K2.6
  37. 90.4 Gemini 3.5 Flash-Lite
  38. 89.6 GPT-5.1
  39. 89.6 Qwen3.7-Max
  40. 88.9 Inkling-Small
  41. 88.5 Qwen3.5 397B-A17B
  42. 88.2 Qwen3.8-2.4T-A95B
  43. 87.4 Kimi K2.5
  44. 87.4 Qwen 3.8 Max
  45. 86.7 Qwen3.5-Plus
  46. 86.7 MiMo-V2.5-Pro
  47. 85.9 Qwen 3.6 Plus
  48. 85.9 GLM-5.1
  49. 85.8 GLM-5.2
  50. 85.2 Gemma 4 31B IT
  51. 85.2 GLM-5.3
  52. 81.5 Claude Sonnet 4.5
  53. 80.7 GPT-4.1
  54. 80.7 GPT-5 mini
  55. 80.7 Qwen3.5-122B-A10B
  56. 80.0 sonar
  57. 80.0 Llama 4 Maverick
  58. 80.0 o4-mini
  59. 80.0 GPT-5.4 Nano
  60. 79.5 MiniMax-M3
  61. 79.3 Muse Glimmer
  62. 77.8 Gemma 4 26B A4B
  63. 77.8 MiMo-V2.5
  64. 76.3 Mistral Large 3
  65. 76.3 MiniMax-M2.7
  66. 74.8 Qwen3.5-27B
  67. 74.1 o3-mini
  68. 74.1 GPT-5 nano
  69. 74.1 Qwen3.5-Flash
  70. 73.9 Qwen3-235B-A22B-Thinking
  71. 73.1 Qwen 3.6 35B-A3B
  72. 72.6 DeepSeek-V3
  73. 71.8 gpt-oss-120b
  74. 71.6 Qwen3.5-35B-A3B
  75. 69.6 Qwen3.7 Flash
  76. 68.2 GPT-4.1 mini
  77. 67.9 Qwen 3.8 27B
  78. 66.9 Qwen3.6 27B
  79. 65.9 Claude Haiku 4.5
  80. 65.7 Gemma 4 12B
  81. 63.9 Doubao-Seed-1.6
  82. 63.7 Mistral Medium 3
  83. 62.2 Mistral Medium 3.1
  84. 62.2 Qwen 3.6 Flash
  85. 61.5 Mistral Small 4
  86. 60.0 Mistral Medium 3.5
  87. 59.3 GLM-4.6V
  88. 55.6 Qwen3-235B-A22B
  89. 54.8 Gemma 3 27B
  90. 54.1 Llama 4 Scout
  91. 54.1 GPT-4.1 nano
  92. 51.9 Qwen3.5-9B
  93. 46.6 glm-4.7-flash
  94. 45.9 Gemma 2 27B
  95. 40.7 Gemma 4 E2B
  96. 40.3 Nemotron 3.5 Lightning 30B-A3B
  97. 40.0 Mistral Small 3
  98. 39.3 Gemma 3 12B
  99. 38.9 Nemotron 3 Nano 30B-A3B
  100. 35.6 Nova 2 Lite
  101. 29.6 Mistral NeMo
  102. 28.9 Phi-4
  103. 26.7 Gemma 3 4B
  104. 22.2 Command R7B
  105. 21.5 Llama 3.2 3B
  106. 20.7 Qwen3-4B
  107. 14.1 Llama 3.2 1B
  108. 11.1 Qwen3-1.7B
  109. 10.4 Qwen3-0.6B
  1. 1 GPT-5.4 medium 97.8%
  2. 2 Gemini 3.7 Flash 97%
  3. 3 Gemini 3 Flash 96.3%
  4. 4 Gemini 3.1 Pro 95.6%
  5. 5 Claude Opus 4.7 95.6%
  6. 6 GPT-5.5 95.6%
  7. 7 DeepSeek-V4-Pro 95.6%
  8. 8 Gemini 3.5 Flash 95.6%
  9. 9 Claude Fable 5 95.6%
  10. 10 Grok 4.5 95.6%
  11. 11 GPT-5.6 Sol 95.6%
  12. 12 DeepSeek-V4-Flash 95.5%
  13. 13 Claude Opus 4.8 94.8%
  14. 14 Kimi K3 94.1%
  15. 15 Grok 4.6 94.1%
  16. 16 GPT-5.6 Terra 93.3%
  17. 17 o3 92.6%
  18. 18 Claude Opus 4.6 92.6%
  19. 19 Grok 4.20 reasoning 92.6%
  20. 20 Inkling 92.6%
  21. 21 Claude Opus 5 92.6%
  22. 22 Gemini 2.5 Pro Jun 2025 91.8%
  23. 23 GPT-5 medium 91.8%
  24. 24 Qwen 3.6 Max (Preview) 91.8%
  25. 25 Muse Spark 1.2 91.8%
  26. 26 Claude Opus 4.5 91.1%
  27. 27 GPT-5.2 medium 91.1%
  28. 28 Gemini 3.1 Flash-Lite 91.1%
  29. 29 GPT-5.4 Mini medium 91.1%
  30. 30 Grok 4.3 91.1%
  31. 31 Claude Sonnet 5 91.1%
  32. 32 GPT-5.6 Luna 91.1%
  33. 33 Gemini 3.6 Flash 91.1%
  34. 34 Gemini 2.5 Flash Sep 2025 90.4%
  35. 35 Claude Sonnet 4.6 90.4%
  36. 36 Kimi K2.6 90.4%
  37. 37 Gemini 3.5 Flash-Lite 90.4%
  38. 38 GPT-5.1 medium 89.6%
  39. 39 Qwen3.7-Max 89.6%
  40. 40 Inkling-Small 88.9%
  41. 41 Qwen3.5 397B-A17B 88.5%
  42. 42 Qwen3.8-2.4T-A95B 88.2%
  43. 43 Kimi K2.5 87.4%
  44. 44 Qwen 3.8 Max 87.4%
  45. 45 Qwen3.5-Plus 86.7%
  46. 46 MiMo-V2.5-Pro 86.7%
  47. 47 Qwen 3.6 Plus 85.9%
  48. 48 GLM-5.1 85.9%
  49. 49 GLM-5.2 85.8%
  50. 50 Gemma 4 31B IT 85.2%
  51. 51 GLM-5.3 85.2%
  52. 52 Claude Sonnet 4.5 81.5%
  53. 53 GPT-4.1 80.7%
  54. 54 GPT-5 mini medium 80.7%
  55. 55 Qwen3.5-122B-A10B 80.7%
  56. 56 sonar 80%
  57. 57 Llama 4 Maverick 80%
  58. 58 o4-mini 80%
  59. 59 GPT-5.4 Nano medium 80%
  60. 60 MiniMax-M3 79.5%
  61. 61 Muse Glimmer 79.3%
  62. 62 Gemma 4 26B A4B 77.8%
  63. 63 MiMo-V2.5 77.8%
  64. 64 Mistral Large 3 76.3%
  65. 65 MiniMax-M2.7 76.3%
  66. 66 Qwen3.5-27B 74.8%
  67. 67 o3-mini 74.1%
  68. 68 GPT-5 nano medium 74.1%
  69. 69 Qwen3.5-Flash 74.1%
  70. 70 Qwen3-235B-A22B-Thinking Jul 2025 73.9%
  71. 71 Qwen 3.6 35B-A3B 73.1%
  72. 72 DeepSeek-V3 Mar 2025 72.6%
  73. 73 gpt-oss-120b 71.8%
  74. 74 Qwen3.5-35B-A3B 71.6%
  75. 75 Qwen3.7 Flash 69.6%
  76. 76 GPT-4.1 mini 68.2%
  77. 77 Qwen 3.8 27B 67.9%
  78. 78 Qwen3.6 27B 66.9%
  79. 79 Claude Haiku 4.5 65.9%
  80. 80 Gemma 4 12B 65.7%
  81. 81 Doubao-Seed-1.6 63.9%
  82. 82 Mistral Medium 3 63.7%
  83. 83 Mistral Medium 3.1 62.2%
  84. 84 Qwen 3.6 Flash 62.2%
  85. 85 Mistral Small 4 61.5%
  86. 86 Mistral Medium 3.5 60%
  87. 87 GLM-4.6V 59.3%
  88. 88 Qwen3-235B-A22B 55.6%
  89. 89 Gemma 3 27B 54.8%
  90. 90 Llama 4 Scout 54.1%
  91. 91 GPT-4.1 nano 54.1%
  92. 92 Qwen3.5-9B 51.9%
  93. 93 glm-4.7-flash 46.6%
  94. 94 Gemma 2 27B 45.9%
  95. 95 Gemma 4 E2B 40.7%
  96. 96 Nemotron 3.5 Lightning 30B-A3B 40.3%
  97. 97 Mistral Small 3 40%
  98. 98 Gemma 3 12B 39.3%
  99. 99 Nemotron 3 Nano 30B-A3B 38.9%
  100. 100 Nova 2 Lite 35.6%
  101. 101 Mistral NeMo 29.6%
  102. 102 Phi-4 28.9%
  103. 103 Gemma 3 4B 26.7%
  104. 104 Command R7B 22.2%
  105. 105 Llama 3.2 3B 21.5%
  106. 106 Qwen3-4B 20.7%
  107. 107 Llama 3.2 1B 14.1%
  108. 108 Qwen3-1.7B 11.1%
  109. 109 Qwen3-0.6B 10.4%

Source: Official Linguistic Knowledge Benchmark (Fab AI) leaderboard

#EntryScoreSettingDateSource
1 GPT-5.4 97.8% medium Mar 2026 Official leaderboard
2 Gemini 3.7 Flash 97% Aug 2026 Official leaderboard
3 Gemini 3 Flash 96.3% Dec 2025 Official leaderboard
4 Claude Opus 4.7 Adaptive Thinking 95.6% Official leaderboard
5 Gemini 3.1 Pro 95.6% Feb 2026 Official leaderboard
6 Claude Opus 4.7 95.6% Apr 2026 Official leaderboard
7 GPT-5.5 95.6% Apr 2026 Official leaderboard
8 DeepSeek-V4-Pro 95.6% Apr 2026 Official leaderboard
9 Gemini 3.5 Flash 95.6% May 2026 Official leaderboard
10 Claude Fable 5 95.6% Jun 2026 Official leaderboard
11 Grok 4.5 95.6% Jul 2026 Official leaderboard
12 GPT-5.6 Sol 95.6% Jul 2026 Official leaderboard
13 DeepSeek-V4-Flash 95.5% Apr 2026 Official leaderboard
14 Claude Opus 4.8 94.8% May 2026 Official leaderboard
15 Kimi K3 94.1% Jul 2026 Official leaderboard
16 Grok 4.6 94.1% Aug 2026 Official leaderboard
17 GPT-5.6 Terra 93.3% Jul 2026 Official leaderboard
18 o3 92.6% Apr 2025 Official leaderboard
19 Claude Opus 4.6 92.6% Feb 2026 Official leaderboard
20 Grok 4.20 92.6% reasoning Feb 2026 Official leaderboard
21 Inkling 92.6% Jul 2026 Official leaderboard
22 Claude Opus 5 92.6% Jul 2026 Official leaderboard
23 Gemini 2.5 Pro Jun 2025 91.8% Jun 2025 Official leaderboard
24 GPT-5 91.8% medium Aug 2025 Official leaderboard
25 Qwen 3.6 Max (Preview) 91.8% Apr 2026 Official leaderboard
26 Muse Spark 1.2 91.8% Aug 2026 Official leaderboard
27 Claude Opus 4.5 91.1% Nov 2025 Official leaderboard
28 GPT-5.2 91.1% medium Dec 2025 Official leaderboard
29 Gemini 3.1 Flash-Lite 91.1% Mar 2026 Official leaderboard
30 GPT-5.4 Mini 91.1% medium Mar 2026 Official leaderboard
31 Grok 4.3 91.1% Apr 2026 Official leaderboard
32 Claude Sonnet 5 91.1% Jun 2026 Official leaderboard
33 GPT-5.6 Luna 91.1% Jul 2026 Official leaderboard
34 Gemini 3.6 Flash 91.1% Jul 2026 Official leaderboard
35 Gemini 2.5 Flash Sep 2025 90.4% Sep 2025 Official leaderboard
36 Claude Sonnet 4.6 90.4% Feb 2026 Official leaderboard
37 Kimi K2.6 90.4% Apr 2026 Official leaderboard
38 Gemini 3.5 Flash-Lite 90.4% Jul 2026 Official leaderboard
39 GPT-5.1 89.6% medium Nov 2025 Official leaderboard
40 Qwen3.7-Max 89.6% May 2026 Official leaderboard
41 Inkling-Small 88.9% Jul 2026 Official leaderboard
42 Qwen3.5 397B-A17B 88.5% Feb 2026 Official leaderboard
43 Qwen3.8-2.4T-A95B 88.2% Aug 2026 Official leaderboard
44 Kimi K2.5 87.4% Jan 2026 Official leaderboard
45 Qwen 3.8 Max 87.4% Aug 2026 Official leaderboard
46 Qwen3.5-Plus 86.7% Feb 2026 Official leaderboard
47 MiMo-V2.5-Pro 86.7% Apr 2026 Official leaderboard
48 Qwen 3.6 Plus 85.9% Mar 2026 Official leaderboard
49 GLM-5.1 85.9% Apr 2026 Official leaderboard
50 GLM-5.2 85.8% Jun 2026 Official leaderboard
51 Gemma 4 31B IT 85.2% Apr 2026 Official leaderboard
52 GLM-5.3 85.2% Aug 2026 Official leaderboard
53 Claude Sonnet 4.5 81.5% Sep 2025 Official leaderboard
54 GPT-4.1 80.7% Apr 2025 Official leaderboard
55 GPT-5 mini 80.7% medium Aug 2025 Official leaderboard
56 Qwen3.5-122B-A10B 80.7% Feb 2026 Official leaderboard
57 sonar 80% Jan 2025 Official leaderboard
58 Llama 4 Maverick 80% Apr 2025 Official leaderboard
59 o4-mini 80% Apr 2025 Official leaderboard
60 GPT-5.4 Nano 80% medium Mar 2026 Official leaderboard
61 MiniMax-M3 79.5% Jun 2026 Official leaderboard
62 Muse Glimmer 79.3% Aug 2026 Official leaderboard
63 Gemma 4 26B A4B 77.8% Apr 2026 Official leaderboard
64 MiMo-V2.5 77.8% Apr 2026 Official leaderboard
65 Mistral Large 3 76.3% Dec 2025 Official leaderboard
66 MiniMax-M2.7 76.3% Mar 2026 Official leaderboard
67 Qwen3.5-27B 74.8% Feb 2026 Official leaderboard
68 o3-mini 74.1% Jan 2025 Official leaderboard
69 GPT-5 nano 74.1% medium Aug 2025 Official leaderboard
70 Qwen3.5-Flash 74.1% Feb 2026 Official leaderboard
71 Qwen3-235B-A22B-Thinking Jul 2025 73.9% Jul 2025 Official leaderboard
72 Qwen 3.6 35B-A3B 73.1% Apr 2026 Official leaderboard
73 DeepSeek-V3 Mar 2025 72.6% Mar 2025 Official leaderboard
74 gpt-oss-120b 71.8% Aug 2025 Official leaderboard
75 Qwen3.5-35B-A3B 71.6% Feb 2026 Official leaderboard
76 Qwen3.7 Flash 69.6% Jul 2026 Official leaderboard
77 GPT-4.1 mini 68.2% Apr 2025 Official leaderboard
78 Qwen 3.8 27B 67.9% Aug 2026 Official leaderboard
79 Qwen3.6 27B 66.9% Apr 2026 Official leaderboard
80 Claude Haiku 4.5 65.9% Oct 2025 Official leaderboard
81 Gemma 4 12B 65.7% Jun 2026 Official leaderboard
82 Doubao-Seed-1.6 63.9% Jun 2025 Official leaderboard
83 Mistral Medium 3 63.7% May 2025 Official leaderboard
84 Mistral Medium 3.1 62.2% Aug 2025 Official leaderboard
85 Qwen 3.6 Flash 62.2% Apr 2026 Official leaderboard
86 Mistral Small 4 61.5% Mar 2026 Official leaderboard
87 Mistral Medium 3.5 60% Apr 2026 Official leaderboard
88 GLM-4.6V 59.3% Dec 2025 Official leaderboard
89 Qwen3-235B-A22B 55.6% Apr 2025 Official leaderboard
90 Gemma 3 27B 54.8% Mar 2025 Official leaderboard
91 Llama 4 Scout 54.1% Apr 2025 Official leaderboard
92 GPT-4.1 nano 54.1% Apr 2025 Official leaderboard
93 Qwen3.5-9B 51.9% Feb 2026 Official leaderboard
94 Command A 50.4% Official leaderboard
95 glm-4.7-flash 46.6% Jan 2026 Official leaderboard
96 Gemma 2 27B 45.9% Jun 2024 Official leaderboard
97 Command-R+ August 42.2% Official leaderboard
98 Gemma 4 E2B 40.7% Mar 2026 Official leaderboard
99 Nemotron 3.5 Lightning 30B-A3B 40.3% Aug 2026 Official leaderboard
100 Mistral Small 3 40% Jan 2025 Official leaderboard
101 Gemma 3 12B 39.3% Mar 2025 Official leaderboard
102 Nemotron 3 Nano 30B-A3B 38.9% Dec 2025 Official leaderboard
103 Nova 2 Lite 35.6% Dec 2025 Official leaderboard
104 Seed 1.6 Flash 33.8% Official leaderboard
105 Mistral NeMo 29.6% Jul 2024 Official leaderboard
106 Phi-4 28.9% Dec 2024 Official leaderboard
107 Gemma 3 4B 26.7% Mar 2025 Official leaderboard
108 Ministral 3 3B 23.7% Official leaderboard
109 Ministral 3 8B 23% Official leaderboard
110 Command R7B 22.2% Dec 2024 Official leaderboard
111 Llama 3.2 3B 21.5% Sep 2024 Official leaderboard
112 Qwen3-4B 20.7% Apr 2025 Official leaderboard
113 Llama 3.2 1B 14.1% Sep 2024 Official leaderboard
114 Qwen3-1.7B 11.1% Apr 2025 Official leaderboard
115 Qwen3-0.6B 10.4% Apr 2025 Official leaderboard
#ModelScoreSettingDateSource
1 GPT-5.4 97.8% medium Mar 2026 Official leaderboard
2 Gemini 3.7 Flash 97% Aug 2026 Official leaderboard
3 Gemini 3 Flash 96.3% Dec 2025 Official leaderboard
4 Gemini 3.1 Pro 95.6% Feb 2026 Official leaderboard
5 Claude Opus 4.7 95.6% Apr 2026 Official leaderboard
6 GPT-5.5 95.6% Apr 2026 Official leaderboard
7 DeepSeek-V4-Pro 95.6% Apr 2026 Official leaderboard
8 Gemini 3.5 Flash 95.6% May 2026 Official leaderboard
9 Claude Fable 5 95.6% Jun 2026 Official leaderboard
10 Grok 4.5 95.6% Jul 2026 Official leaderboard
11 GPT-5.6 Sol 95.6% Jul 2026 Official leaderboard
12 DeepSeek-V4-Flash 95.5% Apr 2026 Official leaderboard
13 Claude Opus 4.8 94.8% May 2026 Official leaderboard
14 Kimi K3 94.1% Jul 2026 Official leaderboard
15 Grok 4.6 94.1% Aug 2026 Official leaderboard
16 GPT-5.6 Terra 93.3% Jul 2026 Official leaderboard
17 o3 92.6% Apr 2025 Official leaderboard
18 Claude Opus 4.6 92.6% Feb 2026 Official leaderboard
19 Grok 4.20 92.6% reasoning Feb 2026 Official leaderboard
20 Inkling 92.6% Jul 2026 Official leaderboard
21 Claude Opus 5 92.6% Jul 2026 Official leaderboard
22 Gemini 2.5 Pro Jun 2025 91.8% Jun 2025 Official leaderboard
23 GPT-5 91.8% medium Aug 2025 Official leaderboard
24 Qwen 3.6 Max (Preview) 91.8% Apr 2026 Official leaderboard
25 Muse Spark 1.2 91.8% Aug 2026 Official leaderboard
26 Claude Opus 4.5 91.1% Nov 2025 Official leaderboard
27 GPT-5.2 91.1% medium Dec 2025 Official leaderboard
28 Gemini 3.1 Flash-Lite 91.1% Mar 2026 Official leaderboard
29 GPT-5.4 Mini 91.1% medium Mar 2026 Official leaderboard
30 Grok 4.3 91.1% Apr 2026 Official leaderboard
31 Claude Sonnet 5 91.1% Jun 2026 Official leaderboard
32 GPT-5.6 Luna 91.1% Jul 2026 Official leaderboard
33 Gemini 3.6 Flash 91.1% Jul 2026 Official leaderboard
34 Gemini 2.5 Flash Sep 2025 90.4% Sep 2025 Official leaderboard
35 Claude Sonnet 4.6 90.4% Feb 2026 Official leaderboard
36 Kimi K2.6 90.4% Apr 2026 Official leaderboard
37 Gemini 3.5 Flash-Lite 90.4% Jul 2026 Official leaderboard
38 GPT-5.1 89.6% medium Nov 2025 Official leaderboard
39 Qwen3.7-Max 89.6% May 2026 Official leaderboard
40 Inkling-Small 88.9% Jul 2026 Official leaderboard
41 Qwen3.5 397B-A17B 88.5% Feb 2026 Official leaderboard
42 Qwen3.8-2.4T-A95B 88.2% Aug 2026 Official leaderboard
43 Kimi K2.5 87.4% Jan 2026 Official leaderboard
44 Qwen 3.8 Max 87.4% Aug 2026 Official leaderboard
45 Qwen3.5-Plus 86.7% Feb 2026 Official leaderboard
46 MiMo-V2.5-Pro 86.7% Apr 2026 Official leaderboard
47 Qwen 3.6 Plus 85.9% Mar 2026 Official leaderboard
48 GLM-5.1 85.9% Apr 2026 Official leaderboard
49 GLM-5.2 85.8% Jun 2026 Official leaderboard
50 Gemma 4 31B IT 85.2% Apr 2026 Official leaderboard
51 GLM-5.3 85.2% Aug 2026 Official leaderboard
52 Claude Sonnet 4.5 81.5% Sep 2025 Official leaderboard
53 GPT-4.1 80.7% Apr 2025 Official leaderboard
54 GPT-5 mini 80.7% medium Aug 2025 Official leaderboard
55 Qwen3.5-122B-A10B 80.7% Feb 2026 Official leaderboard
56 sonar 80% Jan 2025 Official leaderboard
57 Llama 4 Maverick 80% Apr 2025 Official leaderboard
58 o4-mini 80% Apr 2025 Official leaderboard
59 GPT-5.4 Nano 80% medium Mar 2026 Official leaderboard
60 MiniMax-M3 79.5% Jun 2026 Official leaderboard
61 Muse Glimmer 79.3% Aug 2026 Official leaderboard
62 Gemma 4 26B A4B 77.8% Apr 2026 Official leaderboard
63 MiMo-V2.5 77.8% Apr 2026 Official leaderboard
64 Mistral Large 3 76.3% Dec 2025 Official leaderboard
65 MiniMax-M2.7 76.3% Mar 2026 Official leaderboard
66 Qwen3.5-27B 74.8% Feb 2026 Official leaderboard
67 o3-mini 74.1% Jan 2025 Official leaderboard
68 GPT-5 nano 74.1% medium Aug 2025 Official leaderboard
69 Qwen3.5-Flash 74.1% Feb 2026 Official leaderboard
70 Qwen3-235B-A22B-Thinking Jul 2025 73.9% Jul 2025 Official leaderboard
71 Qwen 3.6 35B-A3B 73.1% Apr 2026 Official leaderboard
72 DeepSeek-V3 Mar 2025 72.6% Mar 2025 Official leaderboard
73 gpt-oss-120b 71.8% Aug 2025 Official leaderboard
74 Qwen3.5-35B-A3B 71.6% Feb 2026 Official leaderboard
75 Qwen3.7 Flash 69.6% Jul 2026 Official leaderboard
76 GPT-4.1 mini 68.2% Apr 2025 Official leaderboard
77 Qwen 3.8 27B 67.9% Aug 2026 Official leaderboard
78 Qwen3.6 27B 66.9% Apr 2026 Official leaderboard
79 Claude Haiku 4.5 65.9% Oct 2025 Official leaderboard
80 Gemma 4 12B 65.7% Jun 2026 Official leaderboard
81 Doubao-Seed-1.6 63.9% Jun 2025 Official leaderboard
82 Mistral Medium 3 63.7% May 2025 Official leaderboard
83 Mistral Medium 3.1 62.2% Aug 2025 Official leaderboard
84 Qwen 3.6 Flash 62.2% Apr 2026 Official leaderboard
85 Mistral Small 4 61.5% Mar 2026 Official leaderboard
86 Mistral Medium 3.5 60% Apr 2026 Official leaderboard
87 GLM-4.6V 59.3% Dec 2025 Official leaderboard
88 Qwen3-235B-A22B 55.6% Apr 2025 Official leaderboard
89 Gemma 3 27B 54.8% Mar 2025 Official leaderboard
90 Llama 4 Scout 54.1% Apr 2025 Official leaderboard
91 GPT-4.1 nano 54.1% Apr 2025 Official leaderboard
92 Qwen3.5-9B 51.9% Feb 2026 Official leaderboard
93 glm-4.7-flash 46.6% Jan 2026 Official leaderboard
94 Gemma 2 27B 45.9% Jun 2024 Official leaderboard
95 Gemma 4 E2B 40.7% Mar 2026 Official leaderboard
96 Nemotron 3.5 Lightning 30B-A3B 40.3% Aug 2026 Official leaderboard
97 Mistral Small 3 40% Jan 2025 Official leaderboard
98 Gemma 3 12B 39.3% Mar 2025 Official leaderboard
99 Nemotron 3 Nano 30B-A3B 38.9% Dec 2025 Official leaderboard
100 Nova 2 Lite 35.6% Dec 2025 Official leaderboard
101 Mistral NeMo 29.6% Jul 2024 Official leaderboard
102 Phi-4 28.9% Dec 2024 Official leaderboard
103 Gemma 3 4B 26.7% Mar 2025 Official leaderboard
104 Command R7B 22.2% Dec 2024 Official leaderboard
105 Llama 3.2 3B 21.5% Sep 2024 Official leaderboard
106 Qwen3-4B 20.7% Apr 2025 Official leaderboard
107 Llama 3.2 1B 14.1% Sep 2024 Official leaderboard
108 Qwen3-1.7B 11.1% Apr 2025 Official leaderboard
109 Qwen3-0.6B 10.4% Apr 2025 Official leaderboard

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
GPT-5.4
medium
97.8%
Grok 4.20
reasoning
92.6%
GPT-5.4 Mini
medium
91.1%
GPT-5.4 Nano
medium
80%

Source: Official Linguistic Knowledge Benchmark (Fab AI) leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
GPT-5.4 medium 97.8%
Gemini 3.7 Flash 97%
Gemini 3 Flash 96.3%
Gemini 3.1 Pro 95.6%
Claude Opus 4.7 95.6%
GPT-5.5 95.6%
DeepSeek-V4-Pro 95.6%
Gemini 3.5 Flash 95.6%
Claude Fable 5 95.6%
Grok 4.5 95.6%
GPT-5.6 Sol 95.6%
DeepSeek-V4-Flash 95.5%
DeepSeek-V4-Pro 94.8%
Claude Opus 4.8 94.8%
Kimi K3 94.1%
Grok 4.6 94.1%
GPT-5.6 Terra 93.3%
o3 92.6%
Claude Opus 4.6 92.6%
Grok 4.20 reasoning 92.6%
Inkling 92.6%
Claude Opus 5 92.6%
Gemini 2.5 Pro Jun 2025 91.8%
GPT-5 medium 91.8%
Qwen 3.6 Max (Preview) 91.8%
Muse Spark 1.2 91.8%
Claude Opus 4.5 91.1%
GPT-5.2 medium 91.1%
Gemini 3.1 Flash-Lite 91.1%
GPT-5.4 Mini medium 91.1%
Grok 4.3 91.1%
Claude Sonnet 5 91.1%
GPT-5.6 Luna 91.1%
Gemini 3.6 Flash 91.1%
Gemini 2.5 Flash Sep 2025 90.4%
Claude Sonnet 4.6 90.4%
Kimi K2.6 90.4%
Gemini 3.5 Flash-Lite 90.4%
GPT-5.1 medium 89.6%
Qwen3.7-Max 89.6%
GPT-5.4 88.9%
Inkling-Small 88.9%
Qwen3.5 397B-A17B 88.5%
Qwen3.8-2.4T-A95B 88.2%
Kimi K2.5 87.4%
Gemini 3.1 Flash-Lite 87.4%
Qwen 3.8 Max 87.4%
Qwen3.5-Plus 86.7%
MiMo-V2.5-Pro 86.7%
Qwen 3.6 Plus 85.9%
GLM-5.1 85.9%
GLM-5.2 85.8%
Grok 4.20 85.2%
Gemma 4 31B IT 85.2%
GLM-5.3 85.2%
Claude Sonnet 4.5 81.5%
GPT-4.1 80.7%
GPT-5 mini medium 80.7%
Qwen3.5-122B-A10B 80.7%
sonar 80%
Llama 4 Maverick 80%
o4-mini 80%
GPT-5.4 Mini 80%
GPT-5.4 Nano medium 80%
MiniMax-M3 79.5%
Muse Glimmer 79.3%
Gemma 4 26B A4B 77.8%
MiMo-V2.5 77.8%
Mistral Large 3 76.3%
MiniMax-M2.7 76.3%
Qwen3.5-27B 74.8%
o3-mini 74.1%
GPT-5 nano medium 74.1%
Qwen3.5-Flash 74.1%
Qwen3-235B-A22B-Thinking Jul 2025 73.9%
Qwen 3.6 35B-A3B 73.1%
DeepSeek-V3 Mar 2025 72.6%
gpt-oss-120b 71.8%
Qwen3.5-35B-A3B 71.6%
Qwen3.7 Flash 69.6%
GPT-4.1 mini 68.2%
Qwen 3.8 27B 67.9%
Qwen3.6 27B 66.9%
Claude Haiku 4.5 65.9%
Gemma 4 12B 65.7%
Doubao-Seed-1.6 63.9%
Mistral Medium 3 63.7%
Mistral Medium 3.1 62.2%
Qwen 3.6 Flash 62.2%
Mistral Small 4 61.5%
Mistral Medium 3.5 60%
GLM-4.6V 59.3%
GPT-5.4 Nano 56.3%
Qwen3-235B-A22B 55.6%
Gemma 3 27B 54.8%
Llama 4 Scout 54.1%
GPT-4.1 nano 54.1%
Qwen3.5-9B 51.9%
glm-4.7-flash 46.6%
Gemma 2 27B 45.9%
Gemma 4 E2B 40.7%
Nemotron 3.5 Lightning 30B-A3B 40.3%
Mistral Small 3 40%
Gemma 3 12B 39.3%
Nemotron 3 Nano 30B-A3B 38.9%
Nova 2 Lite 35.6%
Mistral NeMo 29.6%
Phi-4 28.9%
Gemma 3 4B 26.7%
Command R7B 22.2%
Llama 3.2 3B 21.5%
Qwen3-4B 20.7%
Llama 3.2 1B 14.1%
Qwen3-1.7B 11.1%
Qwen3-0.6B 10.4%

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
GPT-5.4
medium
97.8%
Grok 4.20
reasoning
92.6%
GPT-5.4 Mini
medium
91.1%
GPT-5.4 Nano
medium
80%

Source: Official Linguistic Knowledge Benchmark (Fab AI) leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
GPT-5.4 medium 97.8%
Gemini 3.7 Flash 97%
Gemini 3 Flash 96.3%
Gemini 3.1 Pro 95.6%
Claude Opus 4.7 95.6%
GPT-5.5 95.6%
DeepSeek-V4-Pro 95.6%
Gemini 3.5 Flash 95.6%
Claude Fable 5 95.6%
Grok 4.5 95.6%
GPT-5.6 Sol 95.6%
DeepSeek-V4-Flash 95.5%
DeepSeek-V4-Pro 94.8%
Claude Opus 4.8 94.8%
Kimi K3 94.1%
Grok 4.6 94.1%
GPT-5.6 Terra 93.3%
o3 92.6%
Claude Opus 4.6 92.6%
Grok 4.20 reasoning 92.6%
Inkling 92.6%
Claude Opus 5 92.6%
Gemini 2.5 Pro Jun 2025 91.8%
GPT-5 medium 91.8%
Qwen 3.6 Max (Preview) 91.8%
Muse Spark 1.2 91.8%
Claude Opus 4.5 91.1%
GPT-5.2 medium 91.1%
Gemini 3.1 Flash-Lite 91.1%
GPT-5.4 Mini medium 91.1%
Grok 4.3 91.1%
Claude Sonnet 5 91.1%
GPT-5.6 Luna 91.1%
Gemini 3.6 Flash 91.1%
Gemini 2.5 Flash Sep 2025 90.4%
Claude Sonnet 4.6 90.4%
Kimi K2.6 90.4%
Gemini 3.5 Flash-Lite 90.4%
GPT-5.1 medium 89.6%
Qwen3.7-Max 89.6%
GPT-5.4 88.9%
Inkling-Small 88.9%
Qwen3.5 397B-A17B 88.5%
Qwen3.8-2.4T-A95B 88.2%
Kimi K2.5 87.4%
Gemini 3.1 Flash-Lite 87.4%
Qwen 3.8 Max 87.4%
Qwen3.5-Plus 86.7%
MiMo-V2.5-Pro 86.7%
Qwen 3.6 Plus 85.9%
GLM-5.1 85.9%
GLM-5.2 85.8%
Grok 4.20 85.2%
Gemma 4 31B IT 85.2%
GLM-5.3 85.2%
Claude Sonnet 4.5 81.5%
GPT-4.1 80.7%
GPT-5 mini medium 80.7%
Qwen3.5-122B-A10B 80.7%
sonar 80%
Llama 4 Maverick 80%
o4-mini 80%
GPT-5.4 Mini 80%
GPT-5.4 Nano medium 80%
MiniMax-M3 79.5%
Muse Glimmer 79.3%
Gemma 4 26B A4B 77.8%
MiMo-V2.5 77.8%
Mistral Large 3 76.3%
MiniMax-M2.7 76.3%
Qwen3.5-27B 74.8%
o3-mini 74.1%
GPT-5 nano medium 74.1%
Qwen3.5-Flash 74.1%
Qwen3-235B-A22B-Thinking Jul 2025 73.9%
Qwen 3.6 35B-A3B 73.1%
DeepSeek-V3 Mar 2025 72.6%
gpt-oss-120b 71.8%
Qwen3.5-35B-A3B 71.6%
Qwen3.7 Flash 69.6%
GPT-4.1 mini 68.2%
Qwen 3.8 27B 67.9%
Qwen3.6 27B 66.9%
Claude Haiku 4.5 65.9%
Gemma 4 12B 65.7%
Doubao-Seed-1.6 63.9%
Mistral Medium 3 63.7%
Mistral Medium 3.1 62.2%
Qwen 3.6 Flash 62.2%
Mistral Small 4 61.5%
Mistral Medium 3.5 60%
GLM-4.6V 59.3%
GPT-5.4 Nano 56.3%
Qwen3-235B-A22B 55.6%
Gemma 3 27B 54.8%
Llama 4 Scout 54.1%
GPT-4.1 nano 54.1%
Qwen3.5-9B 51.9%
glm-4.7-flash 46.6%
Gemma 2 27B 45.9%
Gemma 4 E2B 40.7%
Nemotron 3.5 Lightning 30B-A3B 40.3%
Mistral Small 3 40%
Gemma 3 12B 39.3%
Nemotron 3 Nano 30B-A3B 38.9%
Nova 2 Lite 35.6%
Mistral NeMo 29.6%
Phi-4 28.9%
Gemma 3 4B 26.7%
Command R7B 22.2%
Llama 3.2 3B 21.5%
Qwen3-4B 20.7%
Llama 3.2 1B 14.1%
Qwen3-1.7B 11.1%
Qwen3-0.6B 10.4%

Number one over time

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

OpenAI
Own leaderboard Passed 85% in Sep 2026, by GPT-5.4
GPT-5.4 97.8%

Source: Official Linguistic Knowledge Benchmark (Fab AI) leaderboard

O3 on the official leaderboard passed 85% in April 2025

Models released before it came out already scored that high. 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 Best so far
0% 20% 40% 60% 80% 100% 2024 2025 2026 Near the ceiling (85%) GPT-5.4 o3 GPT-4.1 Gemma 2 27B 0% 20% 40% 60% 80% 100% 2024 2025 2026 Near the ceiling (85%) GPT-5.4 GPT-4.1

Source: Official Linguistic Knowledge Benchmark (Fab AI) 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 Knowledge

All 332 in Knowledge
  • MMLU-Pro In use Harder 10-option version of MMLU with reasoning-focused questions across 14 subjects. Intern-S2-397B 89.8%
  • MMLU In use 57-subject multiple-choice exam questions from elementary to professional level. Scored by accuracy. Seed1.8 92.3%
  • SuperGPQA In use 26,529 graduate-level multiple-choice questions across 285 disciplines. Accuracy. Qwen 3.6 Max (Preview) 73.9%
  • WideSearch In use 200 bilingual broad info-seeking tasks: agent collects many atomic facts into a table. Scored by success rate and row/item F1. Tencent Hy4 preview 83.9%
  • BBEH micro average In use Models solve harder replacements for each of the 23 BIG-Bench Hard tasks, 4,520 examples in total. The score is the share of all examples answered correctly. Gemma 4 31B IT 74.4%
  • HLE-Diamond In use Models answer 1,000 cleaned expert questions from Humanity's Last Exam, 500 on reasoning and 500 on knowledge, closed book. Graded on exact answers. Claude Fable 5.2 77.9%

Data audit

All 9 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access Public tasks, plus a private test set.
  • Pass Confidence Checked against a primary source.
Scores 3 pass 2 notes
  • Pass Independent runs Scores from the official leaderboard (115 entries).
  • Pass Models scored 109 models we track, plus 6 other entries on its leaderboard.
  • Pass Newest score The newest is from August 2026, for GLM-5.3.
  • Note Frontier models None of the top 10 has a score, which is fine for a retired test. The strongest tested is Claude Opus 5 (#11).
  • Note Headroom Saturated. The best is 97.8%, past 85% since April 2025.
Leaderboard import 3 pass
  • Pass Importer We read the board on fab-ai.org: 115 rows for models we track and 6 other entries at the last read.
  • Pass Last read November 20th, 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