Benchmark Almanac

SimpleQA Verified

Gemini 3.1 Pro 77.5% leads since February 2026
Claude Fable 5.2 94% leads since November 2026
Claude Fable 5.2 94% leads since November 2026
In use

Epoch AI’s run of Claude Fable 5.2, released November 2026, scored 94%. The official leaderboard hasn’t tested it yet.

1,000 human-verified, de-duplicated short-answer factual questions measuring parametric knowledge.

Google fixed one dataset error on March 9th, 2026 and posted paper v2 on March 10th, 2026. Scores stay comparable, so it remains one record.

Made by
Google DeepMind
Released
September 2025
Kind
Short factual questions
Level
University
Scored
Graded by a model
Answers
Short answer
Tasks
All public
Strongest model tested
Claude Fable 5.2 #1 on the capability index
Last in a launch post
September 2026
Last model added
November 2026
Scores refreshed
November 12th, 2026

In the authors’ words

We introduce SimpleQA Verified, a 1,000-prompt benchmark for evaluating Large Language Model (LLM) short-form factuality based on OpenAI's SimpleQA. It addresses critical limitations in OpenAI's benchmark, including noisy and incorrect labels, topical biases, and question redundancy. SimpleQA Verified was created through a rigorous multi-stage filtering process involving de-duplication, topic balancing, and source reconciliation to produce a more reliable and challenging evaluation set, alongside improvements in the autorater prompt. On this new benchmark, Gemini 2.5 Pro achieves a state-of-the-art F1-score of 55.6, outperforming other frontier models, including GPT-5. This work provides the research community with a higher-fidelity tool to track genuine progress in parametric model factuality and to mitigate hallucinations. The benchmark dataset, evaluation code, and leaderboard are available at: https://www.kaggle.com/benchmarks/deepmind/simpleqa-verified.

Abstract of SimpleQA Verified: A Reliable Factuality Benchmark to Measure Parametric Knowledge (arXiv:2509.07968, CC0)
56
Models scored on its own leaderboard
13
Launch posts list it, from 6 labs
20
System cards report it, from 10 labs

Gemini 3.1 Pro leads with 77.5%

Best score per model from the benchmark's own leaderboard, read on November 12th, 2026. Where a model ran with several settings or harnesses, we keep its best.

Claude Fable 5.2 leads with 94% across all independent runs

Best score per model across the official leaderboard, Epoch AI’s own runs and any board Epoch AI copied. Each bar and row names its source.

Claude Fable 5.2 leads with 94%, claims included

Best score per model from every source, including labs’ own numbers and claims Epoch AI collected from reports and papers. Labs pick their own settings, so a claim can beat an independent run of the same model.

56 models. Scroll sideways for the rest.

  1. 77.5 Gemini 3.1 Pro
  2. 75.8 GPT-6 Astra
  3. 75.6 Gemini 4 Argon
  4. 74.6 Gemini 3.8 Flash
  5. 73.8 GPT-6.1 Sol
  6. 73.2 Gemini 3.7 Flash
  7. 72.1 Gemini 3 Pro
  8. 70.4 Gemini 3.5 Flash
  9. 69.2 GPT-5.6 Sol
  10. 64.4 GPT-6 Sol
  11. 63.3 Claude Opus 5
  12. 59.4 Grok 4.7
  13. 53.8 Grok 4.5
  14. 53.7 Qwen3-235B-A22B-Thinking
  15. 52.9 Grok 4.6
  16. 52.3 o3
  17. 51.9 Grok 4
  18. 45.2 Claude Opus 4.6
  19. 44.8 Claude Sonnet 5.5
  20. 44.5 Claude Opus 4.8
  21. 42.3 DeepSeek V4.1 Flash
  22. 41.5 GPT-5.6 Terra
  23. 39.3 Grok 3
  24. 39.1 GPT-5.6 Luna
  25. 39.0 Claude Opus 4.5
  26. 34.9 GPT-5.1
  27. 33.1 Gemini 3.5 Flash-Lite
  28. 32.1 Claude Sonnet 5
  29. 31.0 Claude Opus 4.1
  30. 30.9 Claude Sonnet 4.6
  31. 30.4 GPT-5.4
  32. 29.2 Claude Sonnet 4.5
  33. 28.6 DeepSeek-R1
  34. 28.1 Gemini 2.5 Flash
  35. 27.4 DeepSeek-V3
  36. 24.3 DeepSeek-V3.1
  37. 23.9 Claude Haiku 5.5
  38. 22.3 Mistral Large 2
  39. 21.2 o4-mini
  40. 20.5 Grok-2
  41. 20.2 Grok-3 mini
  42. 19.7 Grok 4 Fast
  43. 17.8 Mixtral 8x22B
  44. 13.1 gpt-oss-120b
  45. 11.2 GPT-3.5 Turbo
  46. 10.7 Gemma 3 27B
  47. 10.7 Gemma 4 31B IT
  48. 9.06 GPT-4o mini
  49. 9.05 Gemma 4 26B A4B
  50. 8.57 Granite 4.0 H Small
  51. 6.82 Ministral 8B
  52. 6.70 Gemma 3 12B
  53. 5.95 gpt-oss-20b
  54. 5.56 Codestral
  55. 5.43 Ministral 3B
  56. 4.60 Gemma 3 4B
  1. 1 Gemini 3.1 Pro 77.5%
  2. 2 GPT-6 Astra 75.8%
  3. 3 Gemini 4 Argon 75.6%
  4. 4 Gemini 3.8 Flash 74.6%
  5. 5 GPT-6.1 Sol 73.8%
  6. 6 Gemini 3.7 Flash 73.2%
  7. 7 Gemini 3 Pro 72.1%
  8. 8 Gemini 3.5 Flash 70.4%
  9. 9 GPT-5.6 Sol 69.2%
  10. 10 GPT-6 Sol 64.4%
  11. 11 Claude Opus 5 63.3%
  12. 12 Grok 4.7 59.4%
  13. 13 Grok 4.5 53.8%
  14. 14 Qwen3-235B-A22B-Thinking Jul 2025 53.7%
  15. 15 Grok 4.6 52.9%
  16. 16 o3 52.3%
  17. 17 Grok 4 51.9%
  18. 18 Claude Opus 4.6 45.2%
  19. 19 Claude Sonnet 5.5 44.8%
  20. 20 Claude Opus 4.8 44.5%
  21. 21 DeepSeek V4.1 Flash 42.3%
  22. 22 GPT-5.6 Terra 41.5%
  23. 23 Grok 3 39.3%
  24. 24 GPT-5.6 Luna 39.1%
  25. 25 Claude Opus 4.5 39%
  26. 26 GPT-5.1 34.9%
  27. 27 Gemini 3.5 Flash-Lite 33.1%
  28. 28 Claude Sonnet 5 32.1%
  29. 29 Claude Opus 4.1 31%
  30. 30 Claude Sonnet 4.6 30.9%
  31. 31 GPT-5.4 30.4%
  32. 32 Claude Sonnet 4.5 thinking 29.2%
  33. 33 DeepSeek-R1 May 2025 28.6%
  34. 34 Gemini 2.5 Flash Jun 2025 28.1%
  35. 35 DeepSeek-V3 Mar 2025 27.4%
  36. 36 DeepSeek-V3.1 24.3%
  37. 37 Claude Haiku 5.5 23.9%
  38. 38 Mistral Large 2 Nov 2024 22.3%
  39. 39 o4-mini 21.2%
  40. 40 Grok-2 Dec 2024 20.5%
  41. 41 Grok-3 mini 20.2%
  42. 42 Grok 4 Fast reasoning 19.7%
  43. 43 Mixtral 8x22B 17.8%
  44. 44 gpt-oss-120b 13.1%
  45. 45 GPT-3.5 Turbo Nov 2023 11.2%
  46. 46 Gemma 3 27B 10.7%
  47. 47 Gemma 4 31B IT 10.7%
  48. 48 GPT-4o mini 9.06%
  49. 49 Gemma 4 26B A4B 9.05%
  50. 50 Granite 4.0 H Small 8.57%
  51. 51 Ministral 8B 6.82%
  52. 52 Gemma 3 12B 6.70%
  53. 53 gpt-oss-20b 5.95%
  54. 54 Codestral 5.56%
  55. 55 Ministral 3B 5.43%
  56. 56 Gemma 3 4B 4.60%

Source: Official SimpleQA Verified leaderboard

119 models. Scroll sideways for the rest.

  1. 94.0 Claude Fable 5.2
  2. 77.7 Gemini 4 Carbon
  3. 77.5 Gemini 3.1 Pro
  4. 76.0 Gemini 4 Argon
  5. 75.8 GPT-6 Astra
  6. 74.6 Gemini 3.8 Flash
  7. 73.9 GPT-6.1 Sol
  8. 73.2 Gemini 3.7 Flash
  9. 72.2 Claude Opus 5.5
  10. 72.1 Gemini 3 Pro
  11. 70.8 Claude Fable 5.1
  12. 70.7 Claude Fable 5
  13. 70.4 Gemini 3.5 Flash
  14. 69.7 GPT-5.6 Sol
  15. 66.8 Gemini 3 Flash
  16. 66.2 Gemini 3.6 Flash
  17. 64.4 GPT-6 Sol
  18. 63.3 Claude Opus 5
  19. 63.0 GPT-5.5
  20. 62.2 Kimi K3.1 Pro
  21. 60.3 Muse Spark 1.2
  22. 59.4 Grok 4.7
  23. 57.8 Muse Spark 1.1
  24. 57.3 GPT-6.1 Terra
  25. 56.6 Muse Spark Open
  26. 55.8 Qwen3.7-Max
  27. 53.8 Grok 4.5
  28. 53.7 Qwen3-235B-A22B-Thinking
  29. 53.0 Claude Opus 4.8
  30. 52.9 DeepSeek V4 Pro 0813
  31. 52.9 Grok 4.6
  32. 52.3 o3
  33. 52.0 Qwen 3.6 Max (Preview)
  34. 51.9 Grok 4
  35. 51.7 Claude Opus 4.7
  36. 50.6 Kimi K3
  37. 50.1 GPT-5
  38. 48.8 Qwen3-Max
  39. 48.0 GPT-5.1
  40. 47.3 Qwen3.8 Max (0902)
  41. 47.0 Claude Opus 4.6
  42. 47.0 DeepSeek-V4-Pro
  43. 46.5 Claude Sonnet 5.5
  44. 46.3 GPT-5.4 Pro
  45. 45.8 Qwen 3.8 Max
  46. 45.7 Claude Opus 4.5
  47. 45.1 GPT-5.4
  48. 44.7 Doubao Seed 2.1 Pro
  49. 44.1 Qwen 3.6 Plus
  50. 43.2 GPT-5.6 Terra
  51. 42.9 DeepSeek V4.1 Flash
  52. 42.5 DeepSeek V4.1 Flash 1017
  53. 41.9 Doubao Seed Evolving
  54. 41.4 GPT-6 Luna
  55. 41.1 o1
  56. 41.0 GPT-5.6 Luna
  57. 41.0 GLM-5.3
  58. 40.3 Inkling
  59. 39.4 Doubao-Seed-2.0 Pro
  60. 39.3 Grok 3
  61. 37.1 GPT-5.2
  62. 36.5 Kimi K2.7 Code
  63. 35.5 Claude Sonnet 4.6
  64. 34.9 Kimi K2.6
  65. 34.3 Kimi K2.5
  66. 34.2 GLM-5.2
  67. 34.0 GLM-5.1
  68. 33.7 Claude Sonnet 5
  69. 33.6 DeepSeek V4 Flash 0731
  70. 33.2 Grok 4.3 Beta
  71. 33.1 Gemini 3.5 Flash-Lite
  72. 32.2 GLM-4.7
  73. 31.5 Doubao Seed 2.0 Lite
  74. 31.1 GPT-4.1
  75. 31.0 Claude Opus 4.1
  76. 30.7 Claude Sonnet 4.5
  77. 30.2 Grok 4.20
  78. 29.7 Doubao Seed 2.1 Turbo
  79. 29.4 GPT-5.4 Mini
  80. 28.6 DeepSeek-R1
  81. 28.1 Gemini 2.5 Flash
  82. 27.4 DeepSeek-V3
  83. 26.0 GPT-4o
  84. 25.4 Qwen 3.5 Plus (hosted 397B-A17B)
  85. 24.3 DeepSeek-V3.1
  86. 23.9 Claude Haiku 5.5
  87. 22.4 Mistral Large 4
  88. 22.3 Mistral Large 2
  89. 21.6 GPT-5 mini
  90. 21.2 o4-mini
  91. 20.5 Grok-2
  92. 20.3 Qwen 3.5 Flash (hosted 35B-A3B)
  93. 20.2 Grok-3 mini
  94. 20.0 Mistral Large 4 Preview
  95. 19.7 Grok 4 Fast
  96. 19.1 Inkling-Small
  97. 17.8 Mixtral 8x22B
  98. 17.4 Doubao-Seed-1.6
  99. 15.9 Qwen 3.6 Flash
  100. 15.3 o3-mini
  101. 13.2 Claude Haiku 4.5
  102. 13.1 gpt-oss-120b
  103. 12.7 GPT-4.1 mini
  104. 12.6 Claude 3 Opus
  105. 11.7 GPT-5 nano
  106. 11.7 GPT-5.4 Nano
  107. 11.2 GPT-3.5 Turbo
  108. 10.7 Gemma 3 27B
  109. 10.7 Gemma 4 31B IT
  110. 9.06 GPT-4o mini
  111. 9.05 Gemma 4 26B A4B
  112. 8.57 Granite 4.0 H Small
  113. 6.82 Ministral 8B
  114. 6.70 Gemma 3 12B
  115. 6.00 GPT-4.1 nano
  116. 5.95 gpt-oss-20b
  117. 5.56 Codestral
  118. 5.43 Ministral 3B
  119. 4.60 Gemma 3 4B
  1. 1 Claude Fable 5.2 max 94%
  2. 2 Gemini 4 Carbon high 77.7%
  3. 3 Gemini 3.1 Pro 77.5%
  4. 4 Gemini 4 Argon high 76%
  5. 5 GPT-6 Astra 75.8%
  6. 6 Gemini 3.8 Flash 74.6%
  7. 7 GPT-6.1 Sol max 73.9%
  8. 8 Gemini 3.7 Flash 73.2%
  9. 9 Claude Opus 5.5 max 72.2%
  10. 10 Gemini 3 Pro 72.1%
  11. 11 Claude Fable 5.1 max 70.8%
  12. 12 Claude Fable 5 xhigh 70.7%
  13. 13 Gemini 3.5 Flash 70.4%
  14. 14 GPT-5.6 Sol max 69.7%
  15. 15 Gemini 3 Flash high 66.8%
  16. 16 Gemini 3.6 Flash high 66.2%
  17. 17 GPT-6 Sol 64.4%
  18. 18 Claude Opus 5 63.3%
  19. 19 GPT-5.5 xhigh 63%
  20. 20 Kimi K3.1 Pro Together, max 62.2%
  21. 21 Muse Spark 1.2 xhigh 60.3%
  22. 22 Grok 4.7 59.4%
  23. 23 Muse Spark 1.1 57.8%
  24. 24 GPT-6.1 Terra max 57.3%
  25. 25 Muse Spark Open xhigh 56.6%
  26. 26 Qwen3.7-Max 55.8%
  27. 27 Grok 4.5 53.8%
  28. 28 Qwen3-235B-A22B-Thinking Jul 2025 53.7%
  29. 29 Claude Opus 4.8 max 53%
  30. 30 DeepSeek V4 Pro 0813 max 52.9%
  31. 31 Grok 4.6 52.9%
  32. 32 o3 52.3%
  33. 33 Qwen 3.6 Max (Preview) 52%
  34. 34 Grok 4 51.9%
  35. 35 Claude Opus 4.7 xhigh 51.7%
  36. 36 Kimi K3 max 50.6%
  37. 37 GPT-5 high 50.1%
  38. 38 Qwen3-Max 48.8%
  39. 39 GPT-5.1 high 48%
  40. 40 Qwen3.8 Max (0902) xhigh 47.3%
  41. 41 Claude Opus 4.6 max 47%
  42. 42 DeepSeek-V4-Pro max 47%
  43. 43 Claude Sonnet 5.5 max 46.5%
  44. 44 GPT-5.4 Pro xhigh 46.3%
  45. 45 Qwen 3.8 Max xhigh 45.8%
  46. 46 Claude Opus 4.5 32k thinking 45.7%
  47. 47 GPT-5.4 xhigh 45.1%
  48. 48 Doubao Seed 2.1 Pro high 44.7%
  49. 49 Qwen 3.6 Plus 2026-04-02 44.1%
  50. 50 GPT-5.6 Terra max 43.2%
  51. 51 DeepSeek V4.1 Flash max 42.9%
  52. 52 DeepSeek V4.1 Flash 1017 max 42.5%
  53. 53 Doubao Seed Evolving Nov 2026, high 41.9%
  54. 54 GPT-6 Luna max 41.4%
  55. 55 o1 high 41.1%
  56. 56 GPT-5.6 Luna max 41%
  57. 57 GLM-5.3 max 41%
  58. 58 Inkling xhigh 40.3%
  59. 59 Doubao-Seed-2.0 Pro high 39.4%
  60. 60 Grok 3 39.3%
  61. 61 GPT-5.2 xhigh 37.1%
  62. 62 Kimi K2.7 Code 36.5%
  63. 63 Claude Sonnet 4.6 high 35.5%
  64. 64 Kimi K2.6 34.9%
  65. 65 Kimi K2.5 34.3%
  66. 66 GLM-5.2 max 34.2%
  67. 67 GLM-5.1 34%
  68. 68 Claude Sonnet 5 max 33.7%
  69. 69 DeepSeek V4 Flash 0731 max 33.6%
  70. 70 Grok 4.3 Beta high 33.2%
  71. 71 Gemini 3.5 Flash-Lite 33.1%
  72. 72 GLM-4.7 32.2%
  73. 73 Doubao Seed 2.0 Lite Apr 2026, high 31.5%
  74. 74 GPT-4.1 31.1%
  75. 75 Claude Opus 4.1 31%
  76. 76 Claude Sonnet 4.5 59k thinking 30.7%
  77. 77 Grok 4.20 30.2%
  78. 78 Doubao Seed 2.1 Turbo high 29.7%
  79. 79 GPT-5.4 Mini high 29.4%
  80. 80 DeepSeek-R1 May 2025 28.6%
  81. 81 Gemini 2.5 Flash Jun 2025 28.1%
  82. 82 DeepSeek-V3 Mar 2025 27.4%
  83. 83 GPT-4o Aug 2024 26%
  84. 84 Qwen 3.5 Plus (hosted 397B-A17B) 25.4%
  85. 85 DeepSeek-V3.1 24.3%
  86. 86 Claude Haiku 5.5 23.9%
  87. 87 Mistral Large 4 high 22.4%
  88. 88 Mistral Large 2 Nov 2024 22.3%
  89. 89 GPT-5 mini high 21.6%
  90. 90 o4-mini 21.2%
  91. 91 Grok-2 Dec 2024 20.5%
  92. 92 Qwen 3.5 Flash (hosted 35B-A3B) 20.3%
  93. 93 Grok-3 mini 20.2%
  94. 94 Mistral Large 4 Preview high 20%
  95. 95 Grok 4 Fast reasoning 19.7%
  96. 96 Inkling-Small xhigh 19.1%
  97. 97 Mixtral 8x22B 17.8%
  98. 98 Doubao-Seed-1.6 thinking 17.4%
  99. 99 Qwen 3.6 Flash 2026-04-16 15.9%
  100. 100 o3-mini high 15.3%
  101. 101 Claude Haiku 4.5 13.2%
  102. 102 gpt-oss-120b 13.1%
  103. 103 GPT-4.1 mini 12.7%
  104. 104 Claude 3 Opus 12.6%
  105. 105 GPT-5 nano high 11.7%
  106. 106 GPT-5.4 Nano high 11.7%
  107. 107 GPT-3.5 Turbo Nov 2023 11.2%
  108. 108 Gemma 3 27B 10.7%
  109. 109 Gemma 4 31B IT 10.7%
  110. 110 GPT-4o mini 9.06%
  111. 111 Gemma 4 26B A4B 9.05%
  112. 112 Granite 4.0 H Small 8.57%
  113. 113 Ministral 8B 6.82%
  114. 114 Gemma 3 12B 6.70%
  115. 115 GPT-4.1 nano 6.00%
  116. 116 gpt-oss-20b 5.95%
  117. 117 Codestral 5.56%
  118. 118 Ministral 3B 5.43%
  119. 119 Gemma 3 4B 4.60%

Sources: Epoch AI runs (CC BY 4.0) and official SimpleQA Verified leaderboard

131 models. Scroll sideways for the rest.

  1. 94.0 Claude Fable 5.2
  2. 78.8 Gemini 4 Argon
  3. 77.7 Gemini 4 Carbon
  4. 77.5 Gemini 3.1 Pro
  5. 75.8 GPT-6 Astra
  6. 74.6 Gemini 3.8 Flash
  7. 73.9 GPT-6.1 Sol
  8. 73.2 Gemini 3.7 Flash
  9. 72.2 Claude Opus 5.5
  10. 72.1 Gemini 3 Pro
  11. 70.8 Muse Spark
  12. 70.8 Claude Fable 5.1
  13. 70.7 Claude Mythos Preview
  14. 70.7 Claude Fable 5
  15. 70.4 Gemini 3.5 Flash
  16. 69.7 GPT-5.6 Sol
  17. 68.7 Gemini 3 Flash
  18. 66.2 Gemini 3.6 Flash
  19. 64.4 GPT-6 Sol
  20. 63.3 Claude Opus 5
  21. 63.0 GPT-5.5
  22. 62.2 Kimi K3.1 Pro
  23. 60.3 Muse Spark 1.2
  24. 59.4 Grok 4.7
  25. 57.9 DeepSeek-V4-Pro
  26. 57.8 Muse Spark 1.1
  27. 57.3 GPT-6.1 Terra
  28. 56.6 Muse Spark Open
  29. 55.8 Qwen3.7-Max
  30. 53.8 Grok 4.5
  31. 53.7 Qwen3-235B-A22B-Thinking
  32. 53.0 Claude Opus 4.8
  33. 52.9 DeepSeek V4 Pro 0813
  34. 52.9 Grok 4.6
  35. 52.3 o3
  36. 52.0 Qwen 3.6 Max (Preview)
  37. 51.9 Grok 4
  38. 51.7 Claude Opus 4.7
  39. 50.6 Kimi K3
  40. 50.1 GPT-5
  41. 48.8 Qwen3-Max
  42. 48.0 GPT-5.1
  43. 47.3 Qwen3.8 Max (0902)
  44. 47.0 Claude Opus 4.6
  45. 46.5 Claude Sonnet 5.5
  46. 46.3 GPT-5.4 Pro
  47. 45.8 Qwen 3.8 Max
  48. 45.7 Claude Opus 4.5
  49. 45.1 GPT-5.4
  50. 44.7 Doubao Seed 2.1 Pro
  51. 44.1 Qwen 3.6 Plus
  52. 43.9 Inkling
  53. 43.3 Gemini 3.1 Flash-Lite
  54. 43.2 GPT-5.6 Terra
  55. 42.9 DeepSeek V4.1 Flash
  56. 42.5 DeepSeek V4.1 Flash 1017
  57. 42.3 DeepSeek-V4.1-Flash Base
  58. 41.9 Doubao Seed Evolving
  59. 41.8 Beam 1.0
  60. 41.4 GPT-6 Luna
  61. 41.1 o1
  62. 41.0 GPT-5.6 Luna
  63. 41.0 GLM-5.3
  64. 39.4 Doubao-Seed-2.0 Pro
  65. 39.3 Grok 3
  66. 37.1 GPT-5.2
  67. 36.9 Kimi K2.5
  68. 36.5 Kimi K2.7 Code
  69. 36.0 Seed2.0 Pro
  70. 35.5 Claude Sonnet 4.6
  71. 34.9 Kimi K2.6
  72. 34.2 GLM-5.2
  73. 34.1 DeepSeek-V4-Flash
  74. 34.0 GLM-5.1
  75. 33.7 Claude Sonnet 5
  76. 33.6 DeepSeek V4 Flash 0731
  77. 33.2 Grok 4.3 Beta
  78. 33.1 Gemini 3.5 Flash-Lite
  79. 32.2 GLM-4.7
  80. 31.5 Doubao Seed 2.0 Lite
  81. 31.5 Ling-2.6-1T
  82. 31.1 GPT-4.1
  83. 31.0 Claude Opus 4.1
  84. 31.0 MAI-Thinking-1
  85. 30.9 Cogito v2.1
  86. 30.7 Claude Sonnet 4.5
  87. 30.2 Grok 4.20
  88. 29.7 Doubao Seed 2.1 Turbo
  89. 29.4 GPT-5.4 Mini
  90. 28.6 DeepSeek-R1
  91. 28.1 Gemini 2.5 Flash
  92. 27.8 Gemini 2.5 Flash
  93. 27.4 DeepSeek-V3
  94. 26.0 GPT-4o
  95. 25.4 Qwen 3.5 Plus (hosted 397B-A17B)
  96. 24.3 DeepSeek-V3.1
  97. 23.9 Claude Haiku 5.5
  98. 22.4 Mistral Large 4
  99. 22.3 Mistral Large 2
  100. 21.6 GPT-5 mini
  101. 21.2 o4-mini
  102. 20.6 Inkling-Small
  103. 20.5 Grok-2
  104. 20.3 Qwen 3.5 Flash (hosted 35B-A3B)
  105. 20.2 Grok-3 mini
  106. 20.0 Mistral Large 4 Preview
  107. 19.7 Grok 4 Fast
  108. 17.8 Mixtral 8x22B
  109. 17.4 Doubao-Seed-1.6
  110. 15.9 Qwen 3.6 Flash
  111. 15.3 o3-mini
  112. 13.2 Claude Haiku 4.5
  113. 13.1 gpt-oss-120b
  114. 12.7 GPT-4.1 mini
  115. 12.6 Claude 3 Opus
  116. 11.7 GPT-5 nano
  117. 11.7 GPT-5.4 Nano
  118. 11.2 GPT-3.5 Turbo
  119. 10.7 Gemma 3 27B
  120. 10.7 Gemma 4 31B IT
  121. 9.60 Gemini 2.5 Flash-Lite
  122. 9.06 GPT-4o mini
  123. 9.05 Gemma 4 26B A4B
  124. 8.57 Granite 4.0 H Small
  125. 6.82 Ministral 8B
  126. 6.70 Gemma 3 12B
  127. 6.00 GPT-4.1 nano
  128. 5.95 gpt-oss-20b
  129. 5.56 Codestral
  130. 5.43 Ministral 3B
  131. 4.60 Gemma 3 4B
  1. 1 Claude Fable 5.2 max 94%
  2. 2 Gemini 4 Argon 78.8%
  3. 3 Gemini 4 Carbon high 77.7%
  4. 4 Gemini 3.1 Pro 77.5%
  5. 5 GPT-6 Astra 75.8%
  6. 6 Gemini 3.8 Flash 74.6%
  7. 7 GPT-6.1 Sol max 73.9%
  8. 8 Gemini 3.7 Flash 73.2%
  9. 9 Claude Opus 5.5 max 72.2%
  10. 10 Gemini 3 Pro 72.1%
  11. 11 Muse Spark 70.8%
  12. 12 Claude Fable 5.1 max 70.8%
  13. 13 Claude Mythos Preview 70.7%
  14. 14 Claude Fable 5 xhigh 70.7%
  15. 15 Gemini 3.5 Flash 70.4%
  16. 16 GPT-5.6 Sol max 69.7%
  17. 17 Gemini 3 Flash 68.7%
  18. 18 Gemini 3.6 Flash high 66.2%
  19. 19 GPT-6 Sol 64.4%
  20. 20 Claude Opus 5 63.3%
  21. 21 GPT-5.5 xhigh 63%
  22. 22 Kimi K3.1 Pro Together, max 62.2%
  23. 23 Muse Spark 1.2 xhigh 60.3%
  24. 24 Grok 4.7 59.4%
  25. 25 DeepSeek-V4-Pro 57.9%
  26. 26 Muse Spark 1.1 57.8%
  27. 27 GPT-6.1 Terra max 57.3%
  28. 28 Muse Spark Open xhigh 56.6%
  29. 29 Qwen3.7-Max 55.8%
  30. 30 Grok 4.5 53.8%
  31. 31 Qwen3-235B-A22B-Thinking Jul 2025 53.7%
  32. 32 Claude Opus 4.8 max 53%
  33. 33 DeepSeek V4 Pro 0813 max 52.9%
  34. 34 Grok 4.6 52.9%
  35. 35 o3 52.3%
  36. 36 Qwen 3.6 Max (Preview) 52%
  37. 37 Grok 4 51.9%
  38. 38 Claude Opus 4.7 xhigh 51.7%
  39. 39 Kimi K3 max 50.6%
  40. 40 GPT-5 high 50.1%
  41. 41 Qwen3-Max 48.8%
  42. 42 GPT-5.1 high 48%
  43. 43 Qwen3.8 Max (0902) xhigh 47.3%
  44. 44 Claude Opus 4.6 max 47%
  45. 45 Claude Sonnet 5.5 max 46.5%
  46. 46 GPT-5.4 Pro xhigh 46.3%
  47. 47 Qwen 3.8 Max xhigh 45.8%
  48. 48 Claude Opus 4.5 32k thinking 45.7%
  49. 49 GPT-5.4 xhigh 45.1%
  50. 50 Doubao Seed 2.1 Pro high 44.7%
  51. 51 Qwen 3.6 Plus 2026-04-02 44.1%
  52. 52 Inkling 43.9%
  53. 53 Gemini 3.1 Flash-Lite 43.3%
  54. 54 GPT-5.6 Terra max 43.2%
  55. 55 DeepSeek V4.1 Flash max 42.9%
  56. 56 DeepSeek V4.1 Flash 1017 max 42.5%
  57. 57 DeepSeek-V4.1-Flash Base 42.3%
  58. 58 Doubao Seed Evolving Nov 2026, high 41.9%
  59. 59 Beam 1.0 41.8%
  60. 60 GPT-6 Luna max 41.4%
  61. 61 o1 high 41.1%
  62. 62 GPT-5.6 Luna max 41%
  63. 63 GLM-5.3 max 41%
  64. 64 Doubao-Seed-2.0 Pro high 39.4%
  65. 65 Grok 3 39.3%
  66. 66 GPT-5.2 xhigh 37.1%
  67. 67 Kimi K2.5 36.9%
  68. 68 Kimi K2.7 Code 36.5%
  69. 69 Seed2.0 Pro 36%
  70. 70 Claude Sonnet 4.6 high 35.5%
  71. 71 Kimi K2.6 34.9%
  72. 72 GLM-5.2 max 34.2%
  73. 73 DeepSeek-V4-Flash 34.1%
  74. 74 GLM-5.1 34%
  75. 75 Claude Sonnet 5 max 33.7%
  76. 76 DeepSeek V4 Flash 0731 max 33.6%
  77. 77 Grok 4.3 Beta high 33.2%
  78. 78 Gemini 3.5 Flash-Lite 33.1%
  79. 79 GLM-4.7 32.2%
  80. 80 Doubao Seed 2.0 Lite Apr 2026, high 31.5%
  81. 81 Ling-2.6-1T 31.5%
  82. 82 GPT-4.1 31.1%
  83. 83 Claude Opus 4.1 31%
  84. 84 MAI-Thinking-1 31%
  85. 85 Cogito v2.1 30.9%
  86. 86 Claude Sonnet 4.5 59k thinking 30.7%
  87. 87 Grok 4.20 30.2%
  88. 88 Doubao Seed 2.1 Turbo high 29.7%
  89. 89 GPT-5.4 Mini high 29.4%
  90. 90 DeepSeek-R1 May 2025 28.6%
  91. 91 Gemini 2.5 Flash Jun 2025 28.1%
  92. 92 Gemini 2.5 Flash Sep 2025 27.8%
  93. 93 DeepSeek-V3 Mar 2025 27.4%
  94. 94 GPT-4o Aug 2024 26%
  95. 95 Qwen 3.5 Plus (hosted 397B-A17B) 25.4%
  96. 96 DeepSeek-V3.1 24.3%
  97. 97 Claude Haiku 5.5 23.9%
  98. 98 Mistral Large 4 high 22.4%
  99. 99 Mistral Large 2 Nov 2024 22.3%
  100. 100 GPT-5 mini high 21.6%
  101. 101 o4-mini 21.2%
  102. 102 Inkling-Small 20.6%
  103. 103 Grok-2 Dec 2024 20.5%
  104. 104 Qwen 3.5 Flash (hosted 35B-A3B) 20.3%
  105. 105 Grok-3 mini 20.2%
  106. 106 Mistral Large 4 Preview high 20%
  107. 107 Grok 4 Fast reasoning 19.7%
  108. 108 Mixtral 8x22B 17.8%
  109. 109 Doubao-Seed-1.6 thinking 17.4%
  110. 110 Qwen 3.6 Flash 2026-04-16 15.9%
  111. 111 o3-mini high 15.3%
  112. 112 Claude Haiku 4.5 13.2%
  113. 113 gpt-oss-120b 13.1%
  114. 114 GPT-4.1 mini 12.7%
  115. 115 Claude 3 Opus 12.6%
  116. 116 GPT-5 nano high 11.7%
  117. 117 GPT-5.4 Nano high 11.7%
  118. 118 GPT-3.5 Turbo Nov 2023 11.2%
  119. 119 Gemma 3 27B 10.7%
  120. 120 Gemma 4 31B IT 10.7%
  121. 121 Gemini 2.5 Flash-Lite Sep 2025 9.60%
  122. 122 GPT-4o mini 9.06%
  123. 123 Gemma 4 26B A4B 9.05%
  124. 124 Granite 4.0 H Small 8.57%
  125. 125 Ministral 8B 6.82%
  126. 126 Gemma 3 12B 6.70%
  127. 127 GPT-4.1 nano 6.00%
  128. 128 gpt-oss-20b 5.95%
  129. 129 Codestral 5.56%
  130. 130 Ministral 3B 5.43%
  131. 131 Gemma 3 4B 4.60%

Sources: Epoch AI runs (CC BY 4.0), labs’ model cards and official SimpleQA Verified leaderboard

#ModelScoreSettingDateSource
1 Gemini 3.1 Pro 77.5% Feb 2026 Official leaderboard
2 GPT-6 Astra 75.8% Sep 2026 Official leaderboard
3 Gemini 4 Argon 75.6% Sep 2026 Official leaderboard
4 Gemini 3.8 Flash 74.6% Sep 2026 Official leaderboard
5 GPT-6.1 Sol 73.8% Sep 2026 Official leaderboard
6 Gemini 3.7 Flash 73.2% Aug 2026 Official leaderboard
7 Gemini 3 Pro 72.1% Nov 2025 Official leaderboard
8 Gemini 3.5 Flash 70.4% May 2026 Official leaderboard
9 GPT-5.6 Sol 69.2% Jul 2026 Official leaderboard
10 GPT-6 Sol 64.4% Sep 2026 Official leaderboard
11 Claude Opus 5 63.3% Jul 2026 Official leaderboard
12 Grok 4.7 59.4% Sep 2026 Official leaderboard
13 Grok 4.5 53.8% Jul 2026 Official leaderboard
14 Qwen3-235B-A22B-Thinking Jul 2025 53.7% Jul 2025 Official leaderboard
15 Grok 4.6 52.9% Aug 2026 Official leaderboard
16 o3 52.3% Apr 2025 Official leaderboard
17 Grok 4 51.9% Jul 2025 Official leaderboard
18 Claude Opus 4.6 45.2% Feb 2026 Official leaderboard
19 Claude Sonnet 5.5 44.8% Sep 2026 Official leaderboard
20 Claude Opus 4.8 44.5% May 2026 Official leaderboard
21 DeepSeek V4.1 Flash 42.3% Sep 2026 Official leaderboard
22 GPT-5.6 Terra 41.5% Jul 2026 Official leaderboard
23 Grok 3 39.3% Apr 2025 Official leaderboard
24 GPT-5.6 Luna 39.1% Jul 2026 Official leaderboard
25 Claude Opus 4.5 39% Nov 2025 Official leaderboard
26 GPT-5.1 34.9% Nov 2025 Official leaderboard
27 Gemini 3.5 Flash-Lite 33.1% Jul 2026 Official leaderboard
28 Claude Sonnet 5 32.1% Jun 2026 Official leaderboard
29 Claude Opus 4.1 31% Aug 2025 Official leaderboard
30 Claude Sonnet 4.6 30.9% Feb 2026 Official leaderboard
31 GPT-5.4 30.4% Mar 2026 Official leaderboard
32 Claude Sonnet 4.5 29.2% thinking Sep 2025 Official leaderboard
33 DeepSeek-R1 May 2025 28.6% May 2025 Official leaderboard
34 Gemini 2.5 Flash Jun 2025 28.1% Jun 2025 Official leaderboard
35 DeepSeek-V3 Mar 2025 27.4% Mar 2025 Official leaderboard
36 DeepSeek-V3.1 24.3% Aug 2025 Official leaderboard
37 Claude Haiku 5.5 23.9% Oct 2026 Official leaderboard
38 Mistral Large 2 Nov 2024 22.3% Nov 2024 Official leaderboard
39 o4-mini 21.2% Apr 2025 Official leaderboard
40 Grok-2 Dec 2024 20.5% Dec 2024 Official leaderboard
41 Grok-3 mini 20.2% Apr 2025 Official leaderboard
42 Grok 4 Fast 19.7% reasoning Sep 2025 Official leaderboard
43 Mixtral 8x22B 17.8% Apr 2024 Official leaderboard
44 gpt-oss-120b 13.1% Aug 2025 Official leaderboard
45 GPT-3.5 Turbo Nov 2023 11.2% Nov 2023 Official leaderboard
46 Gemma 3 27B 10.7% Mar 2025 Official leaderboard
47 Gemma 4 31B IT 10.7% Apr 2026 Official leaderboard
48 GPT-4o mini 9.06% Jul 2024 Official leaderboard
49 Gemma 4 26B A4B 9.05% Apr 2026 Official leaderboard
50 Granite 4.0 H Small 8.57% Oct 2025 Official leaderboard
51 Ministral 8B 6.82% Oct 2024 Official leaderboard
52 Gemma 3 12B 6.70% Mar 2025 Official leaderboard
53 gpt-oss-20b 5.95% Aug 2025 Official leaderboard
54 Codestral 5.56% Jan 2025 Official leaderboard
55 Ministral 3B 5.43% Oct 2024 Official leaderboard
56 Gemma 3 4B 4.60% Mar 2025 Official leaderboard
#ModelScoreSettingDateSource
1 Claude Fable 5.2 94% max Nov 2026 Epoch AI run
2 Gemini 4 Carbon 77.7% high Oct 2026 Epoch AI run
3 Gemini 3.1 Pro 77.5% Feb 2026 Official leaderboard
4 Gemini 4 Argon 76% high Sep 2026 Epoch AI run
5 GPT-6 Astra 75.8% Sep 2026 Official leaderboard
6 Gemini 3.8 Flash 74.6% Sep 2026 Official leaderboard
7 GPT-6.1 Sol 73.9% max Sep 2026 Epoch AI run
8 Gemini 3.7 Flash 73.2% Aug 2026 Official leaderboard
9 Claude Opus 5.5 72.2% max Sep 2026 Epoch AI run
10 Gemini 3 Pro 72.1% Nov 2025 Official leaderboard
11 Claude Fable 5.1 70.8% max Sep 2026 Epoch AI run
12 Claude Fable 5 70.7% xhigh Jun 2026 Epoch AI run
13 Gemini 3.5 Flash 70.4% May 2026 Official leaderboard
14 GPT-5.6 Sol 69.7% max Jul 2026 Epoch AI run
15 Gemini 3 Flash 66.8% high Dec 2025 Epoch AI run
16 Gemini 3.6 Flash 66.2% high Jul 2026 Epoch AI run
17 GPT-6 Sol 64.4% Sep 2026 Official leaderboard
18 Claude Opus 5 63.3% Jul 2026 Official leaderboard
19 GPT-5.5 63% xhigh Apr 2026 Epoch AI run
20 Kimi K3.1 Pro 62.2% Together, max Oct 2026 Epoch AI run
21 Muse Spark 1.2 60.3% xhigh Aug 2026 Epoch AI run
22 Grok 4.7 59.4% Sep 2026 Official leaderboard
23 Muse Spark 1.1 57.8% Jul 2026 Epoch AI run
24 GPT-6.1 Terra 57.3% max Oct 2026 Epoch AI run
25 Muse Spark Open 56.6% xhigh Oct 2026 Epoch AI run
26 Qwen3.7-Max 55.8% May 2026 Epoch AI run
27 Grok 4.5 53.8% Jul 2026 Official leaderboard
28 Qwen3-235B-A22B-Thinking Jul 2025 53.7% Jul 2025 Official leaderboard
29 Claude Opus 4.8 53% max May 2026 Epoch AI run
30 DeepSeek V4 Pro 0813 52.9% max Aug 2026 Epoch AI run
31 Grok 4.6 52.9% Aug 2026 Official leaderboard
32 o3 52.3% Apr 2025 Official leaderboard
33 Qwen 3.6 Max (Preview) 52% Apr 2026 Epoch AI run
34 Grok 4 51.9% Jul 2025 Official leaderboard
35 Claude Opus 4.7 51.7% xhigh Apr 2026 Epoch AI run
36 Kimi K3 50.6% max Jul 2026 Epoch AI run
37 GPT-5 50.1% high Aug 2025 Epoch AI run
38 Qwen3-Max 48.8% Sep 2025 Epoch AI run
39 GPT-5.1 48% high Nov 2025 Epoch AI run
40 Qwen3.8 Max (0902) 47.3% xhigh Sep 2026 Epoch AI run
41 Claude Opus 4.6 47% max Feb 2026 Epoch AI run
42 DeepSeek-V4-Pro 47% max Apr 2026 Epoch AI run
43 Claude Sonnet 5.5 46.5% max Sep 2026 Epoch AI run
44 GPT-5.4 Pro 46.3% xhigh Mar 2026 Epoch AI run
45 Qwen 3.8 Max 45.8% xhigh Aug 2026 Epoch AI run
46 Claude Opus 4.5 45.7% 32k thinking Nov 2025 Epoch AI run
47 GPT-5.4 45.1% xhigh Mar 2026 Epoch AI run
48 Doubao Seed 2.1 Pro 44.7% high Jun 2026 Epoch AI run
49 Qwen 3.6 Plus 44.1% 2026-04-02 Mar 2026 Epoch AI run
50 GPT-5.6 Terra 43.2% max Jul 2026 Epoch AI run
51 DeepSeek V4.1 Flash 42.9% max Sep 2026 Epoch AI run
52 DeepSeek V4.1 Flash 1017 42.5% max Oct 2026 Epoch AI run
53 Doubao Seed Evolving Nov 2026 41.9% high Nov 2026 Epoch AI run
54 GPT-6 Luna 41.4% max Sep 2026 Epoch AI run
55 o1 41.1% high Dec 2024 Epoch AI run
56 GPT-5.6 Luna 41% max Jul 2026 Epoch AI run
57 GLM-5.3 41% max Aug 2026 Epoch AI run
58 Inkling 40.3% xhigh Jul 2026 Epoch AI run
59 Doubao-Seed-2.0 Pro 39.4% high Feb 2026 Epoch AI run
60 Grok 3 39.3% Apr 2025 Official leaderboard
61 GPT-5.2 37.1% xhigh Dec 2025 Epoch AI run
62 Kimi K2.7 Code 36.5% Jun 2026 Epoch AI run
63 Claude Sonnet 4.6 35.5% high Feb 2026 Epoch AI run
64 Kimi K2.6 34.9% Apr 2026 Epoch AI run
65 Kimi K2.5 34.3% Jan 2026 Epoch AI run
66 GLM-5.2 34.2% max Jun 2026 Epoch AI run
67 GLM-5.1 34% Apr 2026 Epoch AI run
68 Claude Sonnet 5 33.7% max Jun 2026 Epoch AI run
69 DeepSeek V4 Flash 0731 33.6% max Jul 2026 Epoch AI run
70 Grok 4.3 Beta 33.2% high Apr 2026 Epoch AI run
71 Gemini 3.5 Flash-Lite 33.1% Jul 2026 Official leaderboard
72 GLM-4.7 32.2% Dec 2025 Epoch AI run
73 Doubao Seed 2.0 Lite Apr 2026 31.5% high Apr 2026 Epoch AI run
74 GPT-4.1 31.1% Apr 2025 Epoch AI run
75 Claude Opus 4.1 31% Aug 2025 Official leaderboard
76 Claude Sonnet 4.5 30.7% 59k thinking Sep 2025 Epoch AI run
77 Grok 4.20 30.2% Feb 2026 Epoch AI run
78 Doubao Seed 2.1 Turbo 29.7% high Jun 2026 Epoch AI run
79 GPT-5.4 Mini 29.4% high Mar 2026 Epoch AI run
80 DeepSeek-R1 May 2025 28.6% May 2025 Official leaderboard
81 Gemini 2.5 Flash Jun 2025 28.1% Jun 2025 Official leaderboard
82 DeepSeek-V3 Mar 2025 27.4% Mar 2025 Official leaderboard
83 GPT-4o Aug 2024 26% Aug 2024 Epoch AI run
84 Qwen 3.5 Plus (hosted 397B-A17B) 25.4% Feb 2026 Epoch AI run
85 DeepSeek-V3.1 24.3% Aug 2025 Official leaderboard
86 Claude Haiku 5.5 23.9% Oct 2026 Official leaderboard
87 Mistral Large 4 22.4% high Oct 2026 Epoch AI run
88 Mistral Large 2 Nov 2024 22.3% Nov 2024 Official leaderboard
89 GPT-5 mini 21.6% high Aug 2025 Epoch AI run
90 o4-mini 21.2% Apr 2025 Official leaderboard
91 Grok-2 Dec 2024 20.5% Dec 2024 Official leaderboard
92 Qwen 3.5 Flash (hosted 35B-A3B) 20.3% Feb 2026 Epoch AI run
93 Grok-3 mini 20.2% Apr 2025 Official leaderboard
94 Mistral Large 4 Preview 20% high Oct 2026 Epoch AI run
95 Grok 4 Fast 19.7% reasoning Sep 2025 Official leaderboard
96 Inkling-Small 19.1% xhigh Jul 2026 Epoch AI run
97 Mixtral 8x22B 17.8% Apr 2024 Official leaderboard
98 Doubao-Seed-1.6 17.4% thinking Jun 2025 Epoch AI run
99 Qwen 3.6 Flash 15.9% 2026-04-16 Apr 2026 Epoch AI run
100 o3-mini 15.3% high Jan 2025 Epoch AI run
101 Claude Haiku 4.5 13.2% Oct 2025 Epoch AI run
102 gpt-oss-120b 13.1% Aug 2025 Official leaderboard
103 GPT-4.1 mini 12.7% Apr 2025 Epoch AI run
104 Claude 3 Opus 12.6% Feb 2024 Epoch AI run
105 GPT-5 nano 11.7% high Aug 2025 Epoch AI run
106 GPT-5.4 Nano 11.7% high Mar 2026 Epoch AI run
107 GPT-3.5 Turbo Nov 2023 11.2% Nov 2023 Official leaderboard
108 Gemma 3 27B 10.7% Mar 2025 Official leaderboard
109 Gemma 4 31B IT 10.7% Apr 2026 Official leaderboard
110 GPT-4o mini 9.06% Jul 2024 Official leaderboard
111 Gemma 4 26B A4B 9.05% Apr 2026 Official leaderboard
112 Granite 4.0 H Small 8.57% Oct 2025 Official leaderboard
113 Ministral 8B 6.82% Oct 2024 Official leaderboard
114 Gemma 3 12B 6.70% Mar 2025 Official leaderboard
115 GPT-4.1 nano 6.00% Apr 2025 Epoch AI run
116 gpt-oss-20b 5.95% Aug 2025 Official leaderboard
117 Codestral 5.56% Jan 2025 Official leaderboard
118 Ministral 3B 5.43% Oct 2024 Official leaderboard
119 Gemma 3 4B 4.60% Mar 2025 Official leaderboard
#ModelScoreSettingDateSource
1 Claude Fable 5.2 94% max Nov 2026 Epoch AI run
2 Gemini 4 Argon 78.8% Oct 2026 Lab model card
3 Gemini 4 Carbon 77.7% high Oct 2026 Epoch AI run
4 Gemini 3.1 Pro 77.5% Feb 2026 Official leaderboard
5 GPT-6 Astra 75.8% Sep 2026 Official leaderboard
6 Gemini 3.8 Flash 74.6% Sep 2026 Official leaderboard
7 GPT-6.1 Sol 73.9% max Sep 2026 Epoch AI run
8 Gemini 3.7 Flash 73.2% Aug 2026 Official leaderboard
9 Claude Opus 5.5 72.2% max Sep 2026 Epoch AI run
10 Gemini 3 Pro 72.1% Nov 2025 Official leaderboard
11 Muse Spark 70.8% Jun 2026 Lab model card
12 Claude Fable 5.1 70.8% max Sep 2026 Epoch AI run
13 Claude Mythos Preview 70.7% Apr 2026 Lab model card
14 Claude Fable 5 70.7% xhigh Jun 2026 Epoch AI run
15 Gemini 3.5 Flash 70.4% May 2026 Official leaderboard
16 GPT-5.6 Sol 69.7% max Jul 2026 Epoch AI run
17 Gemini 3 Flash 68.7% Dec 2025 Lab model card
18 Gemini 3.6 Flash 66.2% high Jul 2026 Epoch AI run
19 GPT-6 Sol 64.4% Sep 2026 Official leaderboard
20 Claude Opus 5 63.3% Jul 2026 Official leaderboard
21 GPT-5.5 63% xhigh Apr 2026 Epoch AI run
22 Kimi K3.1 Pro 62.2% Together, max Oct 2026 Epoch AI run
23 Muse Spark 1.2 60.3% xhigh Aug 2026 Epoch AI run
24 Grok 4.7 59.4% Sep 2026 Official leaderboard
25 DeepSeek-V4-Pro 57.9% Apr 2026 Lab model card
26 Muse Spark 1.1 57.8% Jul 2026 Epoch AI run
27 GPT-6.1 Terra 57.3% max Oct 2026 Epoch AI run
28 Muse Spark Open 56.6% xhigh Oct 2026 Epoch AI run
29 Qwen3.7-Max 55.8% May 2026 Epoch AI run
30 Grok 4.5 53.8% Jul 2026 Official leaderboard
31 Qwen3-235B-A22B-Thinking Jul 2025 53.7% Jul 2025 Official leaderboard
32 Claude Opus 4.8 53% max May 2026 Epoch AI run
33 DeepSeek V4 Pro 0813 52.9% max Aug 2026 Epoch AI run
34 Grok 4.6 52.9% Aug 2026 Official leaderboard
35 o3 52.3% Apr 2025 Official leaderboard
36 Qwen 3.6 Max (Preview) 52% Apr 2026 Epoch AI run
37 Grok 4 51.9% Jul 2025 Official leaderboard
38 Claude Opus 4.7 51.7% xhigh Apr 2026 Epoch AI run
39 Kimi K3 50.6% max Jul 2026 Epoch AI run
40 GPT-5 50.1% high Aug 2025 Epoch AI run
41 Qwen3-Max 48.8% Sep 2025 Epoch AI run
42 GPT-5.1 48% high Nov 2025 Epoch AI run
43 Qwen3.8 Max (0902) 47.3% xhigh Sep 2026 Epoch AI run
44 Claude Opus 4.6 47% max Feb 2026 Epoch AI run
45 Claude Sonnet 5.5 46.5% max Sep 2026 Epoch AI run
46 GPT-5.4 Pro 46.3% xhigh Mar 2026 Epoch AI run
47 Qwen 3.8 Max 45.8% xhigh Aug 2026 Epoch AI run
48 Claude Opus 4.5 45.7% 32k thinking Nov 2025 Epoch AI run
49 GPT-5.4 45.1% xhigh Mar 2026 Epoch AI run
50 Doubao Seed 2.1 Pro 44.7% high Jun 2026 Epoch AI run
51 Qwen 3.6 Plus 44.1% 2026-04-02 Mar 2026 Epoch AI run
52 Inkling 43.9% Jul 2026 Lab model card
53 Gemini 3.1 Flash-Lite 43.3% Mar 2026 Lab model card
54 GPT-5.6 Terra 43.2% max Jul 2026 Epoch AI run
55 DeepSeek V4.1 Flash 42.9% max Sep 2026 Epoch AI run
56 DeepSeek V4.1 Flash 1017 42.5% max Oct 2026 Epoch AI run
57 DeepSeek-V4.1-Flash Base 42.3% Sep 2026 Lab model card
58 Doubao Seed Evolving Nov 2026 41.9% high Nov 2026 Epoch AI run
59 Beam 1.0 41.8% Oct 2026 Lab model card
60 GPT-6 Luna 41.4% max Sep 2026 Epoch AI run
61 o1 41.1% high Dec 2024 Epoch AI run
62 GPT-5.6 Luna 41% max Jul 2026 Epoch AI run
63 GLM-5.3 41% max Aug 2026 Epoch AI run
64 Doubao-Seed-2.0 Pro 39.4% high Feb 2026 Epoch AI run
65 Grok 3 39.3% Apr 2025 Official leaderboard
66 GPT-5.2 37.1% xhigh Dec 2025 Epoch AI run
67 Kimi K2.5 36.9% Jan 2026 Lab model card
68 Kimi K2.7 Code 36.5% Jun 2026 Epoch AI run
69 Seed2.0 Pro 36% Feb 2026 Lab model card
70 Claude Sonnet 4.6 35.5% high Feb 2026 Epoch AI run
71 Kimi K2.6 34.9% Apr 2026 Epoch AI run
72 GLM-5.2 34.2% max Jun 2026 Epoch AI run
73 DeepSeek-V4-Flash 34.1% Apr 2026 Lab model card
74 GLM-5.1 34% Apr 2026 Epoch AI run
75 Claude Sonnet 5 33.7% max Jun 2026 Epoch AI run
76 DeepSeek V4 Flash 0731 33.6% max Jul 2026 Epoch AI run
77 Grok 4.3 Beta 33.2% high Apr 2026 Epoch AI run
78 Gemini 3.5 Flash-Lite 33.1% Jul 2026 Official leaderboard
79 GLM-4.7 32.2% Dec 2025 Epoch AI run
80 Doubao Seed 2.0 Lite Apr 2026 31.5% high Apr 2026 Epoch AI run
81 Ling-2.6-1T 31.5% Jun 2026 Lab model card
82 GPT-4.1 31.1% Apr 2025 Epoch AI run
83 Claude Opus 4.1 31% Aug 2025 Official leaderboard
84 MAI-Thinking-1 31% Jun 2026 Lab model card
85 Cogito v2.1 30.9% Nov 2025 Lab model card
86 Claude Sonnet 4.5 30.7% 59k thinking Sep 2025 Epoch AI run
87 Grok 4.20 30.2% Feb 2026 Epoch AI run
88 Doubao Seed 2.1 Turbo 29.7% high Jun 2026 Epoch AI run
89 GPT-5.4 Mini 29.4% high Mar 2026 Epoch AI run
90 DeepSeek-R1 May 2025 28.6% May 2025 Official leaderboard
91 Gemini 2.5 Flash Jun 2025 28.1% Jun 2025 Official leaderboard
92 Gemini 2.5 Flash Sep 2025 27.8% Sep 2025 Lab model card
93 DeepSeek-V3 Mar 2025 27.4% Mar 2025 Official leaderboard
94 GPT-4o Aug 2024 26% Aug 2024 Epoch AI run
95 Qwen 3.5 Plus (hosted 397B-A17B) 25.4% Feb 2026 Epoch AI run
96 DeepSeek-V3.1 24.3% Aug 2025 Official leaderboard
97 Claude Haiku 5.5 23.9% Oct 2026 Official leaderboard
98 Mistral Large 4 22.4% high Oct 2026 Epoch AI run
99 Mistral Large 2 Nov 2024 22.3% Nov 2024 Official leaderboard
100 GPT-5 mini 21.6% high Aug 2025 Epoch AI run
101 o4-mini 21.2% Apr 2025 Official leaderboard
102 Inkling-Small 20.6% Jul 2026 Lab model card
103 Grok-2 Dec 2024 20.5% Dec 2024 Official leaderboard
104 Qwen 3.5 Flash (hosted 35B-A3B) 20.3% Feb 2026 Epoch AI run
105 Grok-3 mini 20.2% Apr 2025 Official leaderboard
106 Mistral Large 4 Preview 20% high Oct 2026 Epoch AI run
107 Grok 4 Fast 19.7% reasoning Sep 2025 Official leaderboard
108 Mixtral 8x22B 17.8% Apr 2024 Official leaderboard
109 Doubao-Seed-1.6 17.4% thinking Jun 2025 Epoch AI run
110 Qwen 3.6 Flash 15.9% 2026-04-16 Apr 2026 Epoch AI run
111 o3-mini 15.3% high Jan 2025 Epoch AI run
112 Claude Haiku 4.5 13.2% Oct 2025 Epoch AI run
113 gpt-oss-120b 13.1% Aug 2025 Official leaderboard
114 GPT-4.1 mini 12.7% Apr 2025 Epoch AI run
115 Claude 3 Opus 12.6% Feb 2024 Epoch AI run
116 GPT-5 nano 11.7% high Aug 2025 Epoch AI run
117 GPT-5.4 Nano 11.7% high Mar 2026 Epoch AI run
118 GPT-3.5 Turbo Nov 2023 11.2% Nov 2023 Official leaderboard
119 Gemma 3 27B 10.7% Mar 2025 Official leaderboard
120 Gemma 4 31B IT 10.7% Apr 2026 Official leaderboard
121 Gemini 2.5 Flash-Lite Sep 2025 9.60% Sep 2025 Lab model card
122 GPT-4o mini 9.06% Jul 2024 Official leaderboard
123 Gemma 4 26B A4B 9.05% Apr 2026 Official leaderboard
124 Granite 4.0 H Small 8.57% Oct 2025 Official leaderboard
125 Ministral 8B 6.82% Oct 2024 Official leaderboard
126 Gemma 3 12B 6.70% Mar 2025 Official leaderboard
127 GPT-4.1 nano 6.00% Apr 2025 Epoch AI run
128 gpt-oss-20b 5.95% Aug 2025 Official leaderboard
129 Codestral 5.56% Jan 2025 Official leaderboard
130 Ministral 3B 5.43% Oct 2024 Official leaderboard
131 Gemma 3 4B 4.60% Mar 2025 Official leaderboard

Epoch AI has run 94 models. Labs report scores for 22 models in their cards. Switch the scores above to rank them together.

3 models have held number one on the official leaderboard since its release in September 2025 5 models have held number one across all independent runs since its release in September 2025 5 models have held number one with claims included since its release in September 2025

Lab claims count from the day the card came out, independent runs from the model’s release. We draw one row each for the benchmark’s own leaderboard, Epoch AI’s runs and what labs claim in their cards.

Qwen Google DeepMind OpenAI Anthropic
Own leaderboard 3 leaders
Gemini 3.1 Pro 77.5%
Epoch AI runs Passed 85% in Nov 2026, by Claude Fable 5.2
Claude Fable 5.2 94%
Lab claims 3 leaders
Gemini 4 Argon 78.8%

Sources: Official SimpleQA Verified leaderboard, Epoch AI runs (CC BY 4.0) and labs’ model cards

Epoch AI’s run of Claude Fable 5.2 passed 85% in November 2026

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

On the benchmark’s own leaderboard Run by Epoch AI Reported by the lab in its model card Best so far
0% 20% 40% 60% 80% 100% 2024 2025 2026 2027 Near the ceiling (85%) Claude Fable 5.2 Gemini 4 Argon Gemini 3.1 Pro Gemini 3 Pro Qwen3-235B-A22B-Thinking o1 GPT-4o 0% 20% 40% 60% 80% 100% 2024 2025 2026 2027 Near the ceiling (85%) Claude Fable 5.2 Gemini 4 Argon

Sources: Official SimpleQA Verified leaderboard, Epoch AI runs (CC BY 4.0) and labs’ model cards

3 labs put it in a launch post in the last six months

Launch posts from every lab that has listed it, one tick per post. Tall colored ticks include it, and short gray ones leave it out.

Google DeepMind
Deep Cogito
ByteDance Seed
DeepSeek
Microsoft
Thinking Machines
DateLaunch postScoreAs the post writes it
Sep 10, 2026 DeepSeek V4.1 Flash 42.3 SimpleQA-Verified (EM)
Jul 30, 2026 Inkling-Small 20.6 SimpleQA Verified
Jul 15, 2026 Inkling 43.9 SimpleQA Verified
Jun 2, 2026 MAI-Thinking-1 31 SimpleQA Verified
Apr 24, 2026 DeepSeek-V4-Flash 34.1 SimpleQA-Verified (Pass@1), V4-Flash Max
Apr 24, 2026 DeepSeek-V4-Pro 57.9 SimpleQA-Verified (Pass@1)
Mar 3, 2026 Gemini 3.1 Flash-Lite 43.3 SimpleQA Verified
Feb 14, 2026 Seed2.0 Pro 36 SimpleQA Verified
Dec 17, 2025 Gemini 3 Flash 68.7 SimpleQA Verified
Nov 19, 2025 Cogito v2.1 30.9 SimpleQA Verified (f1)
Nov 18, 2025 Gemini 3 Pro 72.1 SimpleQA Verified
Sep 25, 2025 Gemini 2.5 Flash-Lite 9.6 SimpleQA Verified, Thinking
Sep 25, 2025 Gemini 2.5 Flash 27.8 SimpleQA Verified, model card

20 system cards report it, from 10 labs

Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.

DateDocumentScoreAs the document writes itSection
Oct 13, 2026 Gemini 4 Argon system card 78.8 SimpleQA Verified F1 score 78.8; incorrect rate 6.3% (Figures 5.11, 5.12) Hallucination
Oct 5, 2026 Beam 1.0 technical report 41.8 SimpleQA Verified, Table 22 general-capability results Capabilities
Sep 10, 2026 DeepSeek-V4.1-Flash Base model card 42.3 SimpleQA-Verified, EM, 25-shot Capabilities
Sep 10, 2026 DeepSeek V4.1 Flash technical report SimpleQA-Verified (Pass@1) vs reasoning effort, chart only (Figure 12) Hallucination
Jul 30, 2026 Inkling-Small model card 20.6 SimpleQA Verified (5. Evaluations table) Hallucination
Jul 15, 2026 Inkling model card 43.9 SimpleQA Verified (5. Evaluations table, effort 0.99) Hallucination
Jun 13, 2026 Ling-2.6-1T technical report 31.5 SimpleQA-Verified, correct (Table 5) Capabilities
Jun 12, 2026 Muse Spark system card 70.8 SimpleQA Verified F1 (Table 2 labels it Hallucination, Table 21). Correct 67.6, correct given attempted 74.3 Hallucination
Jun 2, 2026 MAI-Thinking-1 technical report 31 SimpleQA Verified, accuracy (Table 12, rounded) Hallucination
May 28, 2026 Claude Opus 4.8 system card SimpleQA Verified net score 0.27 (correct minus incorrect), closed book, printed on chart (6.3.3.1, Figure 6.3.3.1.B) Hallucination
Apr 26, 2026 DeepSeek-V4-Pro technical report 57.9 SimpleQA-Verified (Pass@1), DS-V4-Pro Max (Table 6) Hallucination
Apr 26, 2026 DeepSeek-V4-Flash technical report 34.1 SimpleQA-Verified (Pass@1), DeepSeek-V4-Flash Max (Table 7) Hallucination
Apr 16, 2026 Claude Opus 4.7 system card 44.9 Simple-QA-Verified net score (correct minus incorrect), closed book, printed on chart. Correct 44.9, unsure 38.7, incorrect 16.3 (6.3.3.1, Figures 6.3.3.1.A and B) Hallucination
Apr 7, 2026 Claude Mythos Preview system card 70.7 Simple-QA-Verified correct rate, no tools, value printed on chart; unsure 10.3, incorrect 19.0, net score not printed (4.3.3.1, Figure 4.3.3.1.B) Hallucination
Mar 3, 2026 Gemini 3.1 Flash-Lite model card 43.3 SimpleQA Verified (results table, p.4) Capabilities
Feb 14, 2026 Seed2.0 Pro model card 36 SimpleQA Verified (Table 3) Hallucination
Feb 5, 2026 Claude Opus 4.6 system card Simple-QA-Verified net score (correct minus incorrect), extended thinking, value printed on chart; correct rate 37.7, net 7.8 with thinking plus effort, net 3.6 without thinking (Figure 4.2.B) Hallucination
Jan 30, 2026 Kimi K2.5 technical report 36.9 SimpleQA Verified (Table 4) Hallucination
Nov 24, 2025 Claude Opus 4.5 system card 44.1 Simple-QA-Verified correct rate, extended thinking (Figure 4.1.B; incorrect 35.4%, uncertain 20.5%) Hallucination
Sep 26, 2025 Gemini 2.5 Flash model card 27.8 Factuality SimpleQA Verified (Table, p.4) Capabilities

The lab's number comes out ahead in 7 of 8 pairs that differ

Each score a lab reported for its own model, beside the same model's best score on the leaderboard. Gaps mostly come from different harnesses, effort levels and task subsets. 2 of them match, which usually means one side copied the other. Every benchmark.

Model Lab says Board says Gap
DeepSeek-V4-Pro 57.9% SimpleQA-Verified (Pass@1) 47% Epoch AI run, max +11 pts
Claude Opus 4.7 44.9% Simple-QA-Verified net score (correct minus incorrect), closed book, printed on chart. Correct 44.9, unsure 38.7, incorrect 16.3 (6.3.3.1, Figures 6.3.3.1.A and B) 51.7% Epoch AI run, xhigh −6.8 pts
Claude Opus 4.5 44.1% Simple-QA-Verified correct rate, extended thinking (Figure 4.1.B; incorrect 35.4%, uncertain 20.5%) 39% Leaderboard +5.1 pts
Inkling 43.9% SimpleQA Verified 40.3% Epoch AI run, xhigh +3.6 pts
Gemini 4 Argon 78.8% SimpleQA Verified F1 score 78.8; incorrect rate 6.3% (Figures 5.11, 5.12) 75.6% Leaderboard +3.2 pts
Kimi K2.5 36.9% SimpleQA Verified (Table 4) 34.3% Epoch AI run +2.6 pts
Gemini 3 Flash 68.7% SimpleQA Verified 66.8% Epoch AI run, high +1.9 pts
Inkling-Small 20.6% SimpleQA Verified 19.1% Epoch AI run, xhigh +1.5 pts
DeepSeek V4.1 Flash 42.3% SimpleQA-Verified (EM) 42.3% Leaderboard Same
Gemini 3 Pro 72.1% SimpleQA Verified 72.1% Leaderboard Same

Versions

We switch the top row to each new version when it comes out. Open it for one row per version. An old version stays in play while people still test it, so whoever first takes it past 85% gets credit for closing it out.

OpenAI Perplexity Amazon DeepSeek xAI Qwen Google DeepMind
SimpleQA Verified 12 leaders
Gemini 3.1 Pro 77.5%
SimpleQA Passed 85% in Jan 2025, by Sonar Pro
DeepSeek-V3.2-Exp 97.1%
SimpleQA Verified 3 leaders
Gemini 3.1 Pro 77.5%

Sources: Official leaderboards and labs’ model cards

Replaces SimpleQA

More in Factuality

All 74 in Factuality
  • SimpleVQA In use Models answer 2,025 short factual questions about images across 9 tasks and 9 topics. An LLM judge grades answers against fixed references. Qwen3.7-Plus 81.7%
  • HallusionBench In use Models answer 1,129 yes or no questions about 346 images, built in pairs to expose language hallucination and visual illusion. Scored by accuracy. Keye-VL-2.0-30B-A3B 73.5%
  • Kimi Hallu Bench In use Models answer hard prompts drawn from practical work tasks. Fact-checking agents score the share of claims that are hallucinated. GPT-6 Astra 4.20%
  • Internal adversarially-designed factuality evaluations In use Models answer adversarial questions in nine categories such as history, math and code. Scored on agreement with human ideal responses. GPT-6 Luna 7.60%
  • AA-Omniscience v1.1 In use Models answer 3,000 knowledge questions in six domains without tools and may abstain. The index adds correct answers and subtracts wrong ones. Inkling 2 Small 16.1
  • GaRAGe In use Models write long-form RAG answers to 2,366 questions with human-annotated grounding passages. Scoring checks whether answers use the relevant passages. Nemotron 3.5 Bolt E4B 40.7%