Benchmark Almanac

Human Pathogen Capabilities Test

Claude Opus 4.6 80.7% leads since February 2026
Claude Opus 4.6 80.7% leads since February 2026
Only in system cards

Models answer expert-level questions on human pathogens from SecureBio. Scored as accuracy.

100 text-only questions with a private holdout. 13 PhD virologists scored about 31%. Modeled after VCT. A v2 exists on the SecureBio dashboard.

Made by
SecureBio
Released
August 2025
Kind
Hazardous knowledge
Level
Expert
Scored
Checked automatically
Answers
Multiple choice
Strongest model tested
GPT-6 Astra #3 on the capability index
Last in a system card
October 2026
Last model added
October 2026
Scores refreshed
November 12th, 2026
109
Entries on its own leaderboard, 86 models we track
10
System cards report it, from 3 labs

Claude Opus 4.6 leads with 80.7%

Best score per entry from the benchmark's own leaderboard, read on November 12th, 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders.

Claude Opus 4.6 leads with 80.7%, 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.

109 entries. Scroll sideways for the rest.

  1. 80.7 Claude Opus 4.6
  2. 79.0 Kimi K2.6
  3. 78.6 Claude Opus 5
  4. 77.2 Claude Sonnet 5.5
  5. 76.3 GPT-5.6 Sol
  6. 74.2 Gemini 4 Argon
  7. 69.0 GPT-6 Astra
  8. 68.2 GPT-5.6 Terra
  9. 67.9 Claude Opus 4.5
  10. 67.8 GPT-6 Sol
  11. 66.8 GPT-5.5
  12. 66.5 Claude Sonnet 5
  13. 65.3 GLM-5.2
  14. 64.8 GPT-5.5 Rosalind
  15. 64.7 Gemini 3.7 Flash
  16. 64.0 Gemini 3.6 Flash
  17. 63.9 Gemini 3.1 Pro (Feb 26' Preview)
  18. 63.5 Kimi K3.1 Pro
  19. 63.2 GPT-5.5 pro Codex
  20. 63.1 Claude Opus 4.8
  21. 62.9 GPT-5.1
  22. 62.7 Gemini 3.5 Flash
  23. 62.4 Gemini 3.8 Flash
  24. 61.6 Grok 4.6
  25. 61.1 Gemini 3 Pro (Nov '25 Preview)
  26. 60.7 Beam 1.0
  27. 60.5 DeepSeek V4.1 Flash 1017
  28. 60.1 Claude Opus 4.1
  29. 60.1 GPT-5.2
  30. 59.2 GPT-5
  31. 58.5 Kimi K3
  32. 58.3 o3
  33. 58.0 DeepSeek V4 Pro 0813
  34. 57.8 Qwen3.8-2.4T-A95B
  35. 56.8 Qwen3.7-Max
  36. 56.0 Claude Sonnet 4.5
  37. 56.0 Mistral Large 4
  38. 55.9 GPT-5 Pro
  39. 55.7 GLM-5.3
  40. 54.9 Inkling-Small
  41. 54.5 o4-mini
  42. 54.2 GPT-5 mini
  43. 53.9 Gemini 3 Flash (Dec '25 Preview)
  44. 53.7 Grok 4.7
  45. 53.6 Gemma 4 31B IT
  46. 53.4 GLM-5
  47. 52.7 Nemotron 3 Ultra 550B A55B
  48. 52.4 Kimi K2 Thinking
  49. 52.4 Nemotron 3.5 Super VL
  50. 51.4 Grok 4
  51. 51.1 Inkling
  52. 50.6 Sonar Deep Research
  53. 50.2 Qwen 3.6 Plus
  54. 49.9 Claude 3.5 Sonnet (October 2024)
  55. 49.7 Grok 4.1 Fast
  56. 49.5 Gemini 2.5 Pro (snapshot unspecified)
  57. 49.2 Kimi K2.5
  58. 49.0 Qwen3-235B-A22B-Thinking
  59. 49.0 Qwen 3.8 27B
  60. 48.3 Qwen3.5 397B-A17B
  61. 48.2 o1
  62. 47.5 Nemotron 3 Super 120B A12B
  63. 47.3 GPT-5.4
  64. 47.2 Qwen3-Next-80B-A3B
  65. 46.8 Grok 4 Fast
  66. 45.8 gpt-oss-120b
  67. 45.6 o3-mini
  68. 45.4 GPT-5 nano
  69. 45.3 Qwen3.6 27B
  70. 44.1 DeepSeek-R1
  71. 43.3 GPT-4.5
  72. 42.7 Claude 3.5 Sonnet
  73. 42.0 Claude Sonnet 4
  74. 41.9 Kimi K2 (Jul '25)
  75. 41.7 Gemini 2.5 Flash (snapshot unspecified)
  76. 41.6 DeepSeek-V3.1
  77. 41.6 Claude Haiku 4.5
  78. 41.5 Qwen3-235B-A22B
  79. 41.1 Llama 4 Maverick
  80. 40.9 Grok-3 mini
  81. 40.0 DeepSeek-V3.2
  82. 39.9 Gemini 2.5 Flash (Apr '25 Preview)
  83. 39.0 GPT-4.1 mini
  84. 38.8 DeepSeek-V4-Pro
  85. 38.7 GPT-4.1
  86. 36.2 GPT-4o
  87. 36.2 Nemotron 3 Nano 30B-A3B
  88. 35.3 Kimi K2 (Sep '25)
  89. 35.2 GPT-4.1 nano
  90. 33.8 Mistral Medium 3.5
  91. 33.6 Claude Opus 3
  92. 33.5 Qwen3-Next-80B-A3B-Instruct
  93. 33.4 GPT-4o
  94. 33.0 Grok 3
  95. 29.4 GPT-4 Turbo (snapshot unspecified)
  96. 26.5 Nemotron 3.5 Lightning 30B-A3B
  97. 24.1 Gemini 2.5 Pro (May '25 Preview)
  98. 23.6 Claude 2
  99. 23.2 Mistral Small 3.1
  100. 22.9 GPT-4 (snapshot unspecified)
  101. 22.9 Magistral Medium 1.0
  102. 21.8 Mistral Large 2 (Jul '24)
  103. 21.2 Mistral Large 2.1 (Nov '24)
  104. 20.4 Claude Haiku 3.5
  105. 18.9 Muse Glimmer 30B
  106. 17.9 Magistral Small
  107. 15.1 Mixtral 8x7B
  108. 12.9 GPT-3.5 Turbo (snapshot unspecified)
  109. 12.9 GPT-4o mini
  1. 1 Claude Opus 4.6 max 80.7%
  2. 2 Kimi K2.6 Default 79%
  3. 3 Claude Opus 5 max 78.6%
  4. 4 Claude Sonnet 5.5 max 77.2%
  5. 5 GPT-5.6 Sol max 76.3%
  6. 6 Gemini 4 Argon high 74.2%
  7. 7 GPT-6 Astra max 69%
  8. 8 GPT-5.6 Terra max 68.2%
  9. 9 Claude Opus 4.5 16k 67.9%
  10. 10 GPT-6 Sol max 67.8%
  11. 11 GPT-5.5 xhigh 66.8%
  12. 12 Claude Sonnet 5 max 66.5%
  13. 13 GLM-5.2 max 65.3%
  14. 14 GPT-5.5 Rosalind xhigh 64.8%
  15. 15 Gemini 3.7 Flash high 64.7%
  16. 16 Gemini 3.6 Flash high 64%
  17. 17 Gemini 3.1 Pro (Feb 26' Preview) high 63.9%
  18. 18 Kimi K3.1 Pro max 63.5%
  19. 19 GPT-5.5 pro Codex xhigh 63.2%
  20. 20 Claude Opus 4.8 max 63.1%
  21. 21 GPT-5.1 high 62.9%
  22. 22 Gemini 3.5 Flash high 62.7%
  23. 23 Gemini 3.8 Flash high 62.4%
  24. 24 Grok 4.6 xhigh 61.6%
  25. 25 Gemini 3 Pro (Nov '25 Preview) high 61.1%
  26. 26 Beam 1.0 max 60.7%
  27. 27 DeepSeek V4.1 Flash 1017 max 60.5%
  28. 28 Claude Opus 4.1 16k 60.1%
  29. 29 GPT-5.2 xhigh 60.1%
  30. 30 GPT-5 high 59.2%
  31. 31 Kimi K3 max 58.5%
  32. 32 o3 high 58.3%
  33. 33 DeepSeek V4 Pro 0813 max 58%
  34. 34 Qwen3.8-2.4T-A95B xhigh 57.8%
  35. 35 Qwen3.7-Max 56.8%
  36. 36 Claude Sonnet 4.5 16k 56%
  37. 37 Mistral Large 4 high 56%
  38. 38 GPT-5 Pro high 55.9%
  39. 39 GLM-5.3 max 55.7%
  40. 40 Inkling-Small xhigh 54.9%
  41. 41 o4-mini high 54.5%
  42. 42 GPT-5 mini high 54.2%
  43. 43 Gemini 3 Flash (Dec '25 Preview) high 53.9%
  44. 44 Grok 4.7 xhigh 53.7%
  45. 45 Gemma 4 31B IT Default 53.6%
  46. 46 GLM-5 53.4%
  47. 47 Nemotron 3 Ultra 550B A55B Default 52.7%
  48. 48 Kimi K2 Thinking Default 52.4%
  49. 49 Nemotron 3.5 Super VL Default 52.4%
  50. 50 Grok 4 Default 51.4%
  51. 51 Inkling xhigh 51.1%
  52. 52 Sonar Deep Research high 50.6%
  53. 53 Qwen 3.6 Plus 50.2%
  54. 54 Claude 3.5 Sonnet (October 2024) 49.9%
  55. 55 Grok 4.1 Fast Default 49.7%
  56. 56 Gemini 2.5 Pro (snapshot unspecified) 16k 49.5%
  57. 57 Kimi K2.5 Default 49.2%
  58. 58 Qwen3-235B-A22B-Thinking Jul 2025 49%
  59. 59 Qwen 3.8 27B xhigh 49%
  60. 60 Qwen3.5 397B-A17B 48.3%
  61. 61 o1 high 48.2%
  62. 62 Nemotron 3 Super 120B A12B Default 47.5%
  63. 63 GPT-5.4 xhigh 47.3%
  64. 64 Qwen3-Next-80B-A3B thinking 47.2%
  65. 65 Grok 4 Fast Default 46.8%
  66. 66 gpt-oss-120b high 45.8%
  67. 67 o3-mini high 45.6%
  68. 68 GPT-5 nano high 45.4%
  69. 69 Qwen3.6 27B Default 45.3%
  70. 70 DeepSeek-R1 May 2025, Default 44.1%
  71. 71 GPT-4.5 43.3%
  72. 72 Claude 3.5 Sonnet 42.7%
  73. 73 Claude Sonnet 4 16k 42%
  74. 74 Kimi K2 (Jul '25) 41.9%
  75. 75 Gemini 2.5 Flash (snapshot unspecified) 16k 41.7%
  76. 76 DeepSeek-V3.1 Default 41.6%
  77. 77 Claude Haiku 4.5 16k 41.6%
  78. 78 Qwen3-235B-A22B 41.5%
  79. 79 Llama 4 Maverick 41.1%
  80. 80 Grok-3 mini high 40.9%
  81. 81 DeepSeek-V3.2 Default 40%
  82. 82 Gemini 2.5 Flash (Apr '25 Preview) 16k 39.9%
  83. 83 GPT-4.1 mini 39%
  84. 84 DeepSeek-V4-Pro max 38.8%
  85. 85 GPT-4.1 38.7%
  86. 86 GPT-4o Nov 2024 36.2%
  87. 87 Nemotron 3 Nano 30B-A3B 36.2%
  88. 88 Kimi K2 (Sep '25) 35.3%
  89. 89 GPT-4.1 nano 35.2%
  90. 90 Mistral Medium 3.5 high 33.8%
  91. 91 Claude Opus 3 33.6%
  92. 92 Qwen3-Next-80B-A3B-Instruct 33.5%
  93. 93 GPT-4o Aug 2024 33.4%
  94. 94 Grok 3 33%
  95. 95 GPT-4 Turbo (snapshot unspecified) 29.4%
  96. 96 Nemotron 3.5 Lightning 30B-A3B Default 26.5%
  97. 97 Gemini 2.5 Pro (May '25 Preview) 16k 24.1%
  98. 98 Claude 2 23.6%
  99. 99 Mistral Small 3.1 23.2%
  100. 100 GPT-4 (snapshot unspecified) 22.9%
  101. 101 Magistral Medium 1.0 Default 22.9%
  102. 102 Mistral Large 2 (Jul '24) 21.8%
  103. 103 Mistral Large 2.1 (Nov '24) 21.2%
  104. 104 Claude Haiku 3.5 20.4%
  105. 105 Muse Glimmer 30B xhigh 18.9%
  106. 106 Magistral Small Default 17.9%
  107. 107 Mixtral 8x7B 15.1%
  108. 108 GPT-3.5 Turbo (snapshot unspecified) 12.9%
  109. 109 GPT-4o mini 12.9%

Source: Official Human Pathogen Capabilities Test leaderboard

91 models. Scroll sideways for the rest.

  1. 80.7 Claude Opus 4.6
  2. 79.0 Kimi K2.6
  3. 78.6 Claude Opus 5
  4. 77.2 Claude Sonnet 5.5
  5. 76.3 GPT-5.6 Sol
  6. 74.2 Gemini 4 Argon
  7. 69.0 GPT-6 Astra
  8. 68.2 GPT-5.6 Terra
  9. 67.9 Claude Opus 4.5
  10. 67.8 GPT-6 Sol
  11. 66.8 GPT-5.5
  12. 66.5 Claude Sonnet 5
  13. 66.5 Muse Wonder Full
  14. 65.3 GLM-5.2
  15. 64.7 Gemini 3.7 Flash
  16. 64.0 Gemini 3.6 Flash
  17. 63.5 Kimi K3.1 Pro
  18. 63.1 Claude Opus 4.8
  19. 62.9 GPT-5.1
  20. 62.7 Gemini 3.5 Flash
  21. 62.7 Muse Spark 1.3
  22. 62.4 Gemini 3.8 Flash
  23. 61.9 Muse Spark 1.1
  24. 61.6 Grok 4.6
  25. 61.0 Beam 1.0
  26. 60.5 DeepSeek V4.1 Flash 1017
  27. 60.1 Claude Opus 4.1
  28. 60.1 GPT-5.2
  29. 59.2 GPT-5
  30. 58.5 Kimi K3
  31. 58.3 o3
  32. 58.0 DeepSeek V4 Pro 0813
  33. 57.8 Qwen3.8-2.4T-A95B
  34. 56.8 Qwen3.7-Max
  35. 56.0 Claude Sonnet 4.5
  36. 56.0 Mistral Large 4
  37. 55.9 GPT-5 Pro
  38. 55.7 Muse Spark
  39. 55.7 GLM-5.3
  40. 54.9 Inkling-Small
  41. 54.5 o4-mini
  42. 54.2 GPT-5 mini
  43. 53.7 Grok 4.7
  44. 53.6 Gemma 4 31B IT
  45. 53.4 GLM-5
  46. 52.4 Kimi K2 Thinking
  47. 52.4 Nemotron 3.5 Super VL
  48. 51.4 Grok 4
  49. 51.1 Inkling
  50. 50.2 Qwen 3.6 Plus
  51. 49.9 Claude 3.5 Sonnet (October 2024)
  52. 49.7 Grok 4.1 Fast
  53. 49.2 Kimi K2.5
  54. 49.0 Qwen3-235B-A22B-Thinking
  55. 49.0 Qwen 3.8 27B
  56. 48.3 Qwen3.5 397B-A17B
  57. 48.2 o1
  58. 47.3 GPT-5.4
  59. 47.2 Qwen3-Next-80B-A3B
  60. 46.8 Grok 4 Fast
  61. 45.8 gpt-oss-120b
  62. 45.6 o3-mini
  63. 45.4 GPT-5 nano
  64. 45.3 Qwen3.6 27B
  65. 44.1 DeepSeek-R1
  66. 43.3 GPT-4.5
  67. 42.7 Claude 3.5 Sonnet
  68. 42.0 Claude Sonnet 4
  69. 41.6 DeepSeek-V3.1
  70. 41.6 Claude Haiku 4.5
  71. 41.5 Qwen3-235B-A22B
  72. 41.1 Llama 4 Maverick
  73. 40.9 Grok-3 mini
  74. 40.0 DeepSeek-V3.2
  75. 39.0 GPT-4.1 mini
  76. 38.8 DeepSeek-V4-Pro
  77. 38.7 GPT-4.1
  78. 36.2 GPT-4o
  79. 36.2 Nemotron 3 Nano 30B-A3B
  80. 35.2 GPT-4.1 nano
  81. 33.8 Mistral Medium 3.5
  82. 33.5 Qwen3-Next-80B-A3B-Instruct
  83. 33.4 GPT-4o
  84. 33.0 Grok 3
  85. 31.2 Code World Model (CWM)
  86. 26.5 Nemotron 3.5 Lightning 30B-A3B
  87. 23.6 Claude 2
  88. 23.2 Mistral Small 3.1
  89. 22.9 Magistral Medium 1.0
  90. 15.1 Mixtral 8x7B
  91. 12.9 GPT-4o mini
  1. 1 Claude Opus 4.6 max 80.7%
  2. 2 Kimi K2.6 Default 79%
  3. 3 Claude Opus 5 max 78.6%
  4. 4 Claude Sonnet 5.5 max 77.2%
  5. 5 GPT-5.6 Sol max 76.3%
  6. 6 Gemini 4 Argon high 74.2%
  7. 7 GPT-6 Astra max 69%
  8. 8 GPT-5.6 Terra max 68.2%
  9. 9 Claude Opus 4.5 16k 67.9%
  10. 10 GPT-6 Sol max 67.8%
  11. 11 GPT-5.5 xhigh 66.8%
  12. 12 Claude Sonnet 5 max 66.5%
  13. 13 Muse Wonder Full 66.5%
  14. 14 GLM-5.2 max 65.3%
  15. 15 Gemini 3.7 Flash high 64.7%
  16. 16 Gemini 3.6 Flash high 64%
  17. 17 Kimi K3.1 Pro max 63.5%
  18. 18 Claude Opus 4.8 max 63.1%
  19. 19 GPT-5.1 high 62.9%
  20. 20 Gemini 3.5 Flash high 62.7%
  21. 21 Muse Spark 1.3 62.7%
  22. 22 Gemini 3.8 Flash high 62.4%
  23. 23 Muse Spark 1.1 61.9%
  24. 24 Grok 4.6 xhigh 61.6%
  25. 25 Beam 1.0 61%
  26. 26 DeepSeek V4.1 Flash 1017 max 60.5%
  27. 27 Claude Opus 4.1 16k 60.1%
  28. 28 GPT-5.2 xhigh 60.1%
  29. 29 GPT-5 high 59.2%
  30. 30 Kimi K3 max 58.5%
  31. 31 o3 high 58.3%
  32. 32 DeepSeek V4 Pro 0813 max 58%
  33. 33 Qwen3.8-2.4T-A95B xhigh 57.8%
  34. 34 Qwen3.7-Max 56.8%
  35. 35 Claude Sonnet 4.5 16k 56%
  36. 36 Mistral Large 4 high 56%
  37. 37 GPT-5 Pro high 55.9%
  38. 38 Muse Spark 55.7%
  39. 39 GLM-5.3 max 55.7%
  40. 40 Inkling-Small xhigh 54.9%
  41. 41 o4-mini high 54.5%
  42. 42 GPT-5 mini high 54.2%
  43. 43 Grok 4.7 xhigh 53.7%
  44. 44 Gemma 4 31B IT Default 53.6%
  45. 45 GLM-5 53.4%
  46. 46 Kimi K2 Thinking Default 52.4%
  47. 47 Nemotron 3.5 Super VL Default 52.4%
  48. 48 Grok 4 Default 51.4%
  49. 49 Inkling xhigh 51.1%
  50. 50 Qwen 3.6 Plus 50.2%
  51. 51 Claude 3.5 Sonnet (October 2024) 49.9%
  52. 52 Grok 4.1 Fast Default 49.7%
  53. 53 Kimi K2.5 Default 49.2%
  54. 54 Qwen3-235B-A22B-Thinking Jul 2025 49%
  55. 55 Qwen 3.8 27B xhigh 49%
  56. 56 Qwen3.5 397B-A17B 48.3%
  57. 57 o1 high 48.2%
  58. 58 GPT-5.4 xhigh 47.3%
  59. 59 Qwen3-Next-80B-A3B thinking 47.2%
  60. 60 Grok 4 Fast Default 46.8%
  61. 61 gpt-oss-120b high 45.8%
  62. 62 o3-mini high 45.6%
  63. 63 GPT-5 nano high 45.4%
  64. 64 Qwen3.6 27B Default 45.3%
  65. 65 DeepSeek-R1 May 2025, Default 44.1%
  66. 66 GPT-4.5 43.3%
  67. 67 Claude 3.5 Sonnet 42.7%
  68. 68 Claude Sonnet 4 16k 42%
  69. 69 DeepSeek-V3.1 Default 41.6%
  70. 70 Claude Haiku 4.5 16k 41.6%
  71. 71 Qwen3-235B-A22B 41.5%
  72. 72 Llama 4 Maverick 41.1%
  73. 73 Grok-3 mini high 40.9%
  74. 74 DeepSeek-V3.2 Default 40%
  75. 75 GPT-4.1 mini 39%
  76. 76 DeepSeek-V4-Pro max 38.8%
  77. 77 GPT-4.1 38.7%
  78. 78 GPT-4o Nov 2024 36.2%
  79. 79 Nemotron 3 Nano 30B-A3B 36.2%
  80. 80 GPT-4.1 nano 35.2%
  81. 81 Mistral Medium 3.5 high 33.8%
  82. 82 Qwen3-Next-80B-A3B-Instruct 33.5%
  83. 83 GPT-4o Aug 2024 33.4%
  84. 84 Grok 3 33%
  85. 85 Code World Model (CWM) 31.2%
  86. 86 Nemotron 3.5 Lightning 30B-A3B Default 26.5%
  87. 87 Claude 2 23.6%
  88. 88 Mistral Small 3.1 23.2%
  89. 89 Magistral Medium 1.0 Default 22.9%
  90. 90 Mixtral 8x7B 15.1%
  91. 91 GPT-4o mini 12.9%

Sources: Official Human Pathogen Capabilities Test leaderboard and labs’ model cards

#EntryScoreSettingDateSource
1 Claude Opus 4.6 80.7% max Feb 2026 Official leaderboard
2 Kimi K2.6 79% Default Apr 2026 Official leaderboard
3 Claude Opus 5 78.6% max Jul 2026 Official leaderboard
4 Claude Sonnet 5.5 77.2% max Sep 2026 Official leaderboard
5 GPT-5.6 Sol 76.3% max Jul 2026 Official leaderboard
6 Gemini 4 Argon 74.2% high Sep 2026 Official leaderboard
7 GPT-6 Astra 69% max Sep 2026 Official leaderboard
8 GPT-5.6 Terra 68.2% max Jul 2026 Official leaderboard
9 Claude Opus 4.5 67.9% 16k Nov 2025 Official leaderboard
10 GPT-6 Sol 67.8% max Sep 2026 Official leaderboard
11 GPT-5.5 66.8% xhigh Apr 2026 Official leaderboard
12 Claude Sonnet 5 66.5% max Jun 2026 Official leaderboard
13 GLM-5.2 65.3% max Jun 2026 Official leaderboard
14 GPT-5.5 Rosalind 64.8% xhigh Official leaderboard
15 Gemini 3.7 Flash 64.7% high Aug 2026 Official leaderboard
16 Gemini 3.6 Flash 64% high Jul 2026 Official leaderboard
17 Gemini 3.1 Pro (Feb 26' Preview) 63.9% high Official leaderboard
18 Kimi K3.1 Pro 63.5% max Oct 2026 Official leaderboard
19 GPT-5.5 pro Codex 63.2% xhigh Official leaderboard
20 Claude Opus 4.8 63.1% max May 2026 Official leaderboard
21 GPT-5.1 62.9% high Nov 2025 Official leaderboard
22 Gemini 3.5 Flash 62.7% high May 2026 Official leaderboard
23 Gemini 3.8 Flash 62.4% high Sep 2026 Official leaderboard
24 Grok 4.6 61.6% xhigh Aug 2026 Official leaderboard
25 Gemini 3 Pro (Nov '25 Preview) 61.1% high Official leaderboard
26 Beam 1.0 60.7% max Oct 2026 Official leaderboard
27 DeepSeek V4.1 Flash 1017 60.5% max Oct 2026 Official leaderboard
28 Claude Opus 4.1 60.1% 16k Aug 2025 Official leaderboard
29 GPT-5.2 60.1% xhigh Dec 2025 Official leaderboard
30 GPT-5 59.2% high Aug 2025 Official leaderboard
31 Kimi K3 58.5% max Jul 2026 Official leaderboard
32 o3 58.3% high Apr 2025 Official leaderboard
33 DeepSeek V4 Pro 0813 58% max Aug 2026 Official leaderboard
34 Qwen3.8-2.4T-A95B 57.8% xhigh Aug 2026 Official leaderboard
35 Qwen3.7-Max 56.8% May 2026 Official leaderboard
36 Claude Sonnet 4.5 56% 16k Sep 2025 Official leaderboard
37 Mistral Large 4 56% high Oct 2026 Official leaderboard
38 GPT-5 Pro 55.9% high Oct 2025 Official leaderboard
39 GLM-5.3 55.7% max Aug 2026 Official leaderboard
40 Inkling-Small 54.9% xhigh Jul 2026 Official leaderboard
41 o4-mini 54.5% high Apr 2025 Official leaderboard
42 GPT-5 mini 54.2% high Aug 2025 Official leaderboard
43 Gemini 3 Flash (Dec '25 Preview) 53.9% high Official leaderboard
44 Grok 4.7 53.7% xhigh Sep 2026 Official leaderboard
45 Gemma 4 31B IT 53.6% Default Apr 2026 Official leaderboard
46 GLM-5 53.4% Feb 2026 Official leaderboard
47 Nemotron 3 Ultra 550B A55B 52.7% Default Official leaderboard
48 Kimi K2 Thinking 52.4% Default Nov 2025 Official leaderboard
49 Nemotron 3.5 Super VL 52.4% Default Oct 2026 Official leaderboard
50 Grok 4 51.4% Default Jul 2025 Official leaderboard
51 Inkling 51.1% xhigh Jul 2026 Official leaderboard
52 Sonar Deep Research 50.6% high Official leaderboard
53 Qwen 3.6 Plus 50.2% Mar 2026 Official leaderboard
54 Claude 3.5 Sonnet (October 2024) 49.9% Oct 2024 Official leaderboard
55 Grok 4.1 Fast 49.7% Default Nov 2025 Official leaderboard
56 Gemini 2.5 Pro (snapshot unspecified) 49.5% 16k Official leaderboard
57 Kimi K2.5 49.2% Default Jan 2026 Official leaderboard
58 Qwen3-235B-A22B-Thinking Jul 2025 49% Jul 2025 Official leaderboard
59 Qwen 3.8 27B 49% xhigh Aug 2026 Official leaderboard
60 Qwen3.5 397B-A17B 48.3% Feb 2026 Official leaderboard
61 o1 48.2% high Dec 2024 Official leaderboard
62 Nemotron 3 Super 120B A12B 47.5% Default Official leaderboard
63 GPT-5.4 47.3% xhigh Mar 2026 Official leaderboard
64 Qwen3-Next-80B-A3B 47.2% thinking Sep 2025 Official leaderboard
65 Grok 4 Fast 46.8% Default Sep 2025 Official leaderboard
66 gpt-oss-120b 45.8% high Aug 2025 Official leaderboard
67 o3-mini 45.6% high Jan 2025 Official leaderboard
68 GPT-5 nano 45.4% high Aug 2025 Official leaderboard
69 Qwen3.6 27B 45.3% Default Apr 2026 Official leaderboard
70 DeepSeek-R1 May 2025 44.1% Default May 2025 Official leaderboard
71 GPT-4.5 43.3% Feb 2025 Official leaderboard
72 Claude 3.5 Sonnet 42.7% Jun 2024 Official leaderboard
73 Claude Sonnet 4 42% 16k May 2025 Official leaderboard
74 Kimi K2 (Jul '25) 41.9% Official leaderboard
75 Gemini 2.5 Flash (snapshot unspecified) 41.7% 16k Official leaderboard
76 DeepSeek-V3.1 41.6% Default Aug 2025 Official leaderboard
77 Claude Haiku 4.5 41.6% 16k Oct 2025 Official leaderboard
78 Qwen3-235B-A22B 41.5% Apr 2025 Official leaderboard
79 Llama 4 Maverick 41.1% Apr 2025 Official leaderboard
80 Grok-3 mini 40.9% high Apr 2025 Official leaderboard
81 DeepSeek-V3.2 40% Default Dec 2025 Official leaderboard
82 Gemini 2.5 Flash (Apr '25 Preview) 39.9% 16k Official leaderboard
83 GPT-4.1 mini 39% Apr 2025 Official leaderboard
84 DeepSeek-V4-Pro 38.8% max Apr 2026 Official leaderboard
85 GPT-4.1 38.7% Apr 2025 Official leaderboard
86 GPT-4o Nov 2024 36.2% Nov 2024 Official leaderboard
87 Nemotron 3 Nano 30B-A3B 36.2% Dec 2025 Official leaderboard
88 Kimi K2 (Sep '25) 35.3% Official leaderboard
89 GPT-4.1 nano 35.2% Apr 2025 Official leaderboard
90 Mistral Medium 3.5 33.8% high Apr 2026 Official leaderboard
91 Claude Opus 3 33.6% Official leaderboard
92 Qwen3-Next-80B-A3B-Instruct 33.5% Sep 2025 Official leaderboard
93 GPT-4o Aug 2024 33.4% Aug 2024 Official leaderboard
94 Grok 3 33% Apr 2025 Official leaderboard
95 GPT-4 Turbo (snapshot unspecified) 29.4% Official leaderboard
96 Nemotron 3.5 Lightning 30B-A3B 26.5% Default Aug 2026 Official leaderboard
97 Gemini 2.5 Pro (May '25 Preview) 24.1% 16k Official leaderboard
98 Claude 2 23.6% Jul 2023 Official leaderboard
99 Mistral Small 3.1 23.2% Mar 2025 Official leaderboard
100 GPT-4 (snapshot unspecified) 22.9% Official leaderboard
101 Magistral Medium 1.0 22.9% Default Jun 2025 Official leaderboard
102 Mistral Large 2 (Jul '24) 21.8% Official leaderboard
103 Mistral Large 2.1 (Nov '24) 21.2% Official leaderboard
104 Claude Haiku 3.5 20.4% Official leaderboard
105 Muse Glimmer 30B 18.9% xhigh Official leaderboard
106 Magistral Small 17.9% Default Official leaderboard
107 Mixtral 8x7B 15.1% Dec 2023 Official leaderboard
108 GPT-3.5 Turbo (snapshot unspecified) 12.9% Official leaderboard
109 GPT-4o mini 12.9% Jul 2024 Official leaderboard
#ModelScoreSettingDateSource
1 Claude Opus 4.6 80.7% max Feb 2026 Official leaderboard
2 Kimi K2.6 79% Default Apr 2026 Official leaderboard
3 Claude Opus 5 78.6% max Jul 2026 Official leaderboard
4 Claude Sonnet 5.5 77.2% max Sep 2026 Official leaderboard
5 GPT-5.6 Sol 76.3% max Jul 2026 Official leaderboard
6 Gemini 4 Argon 74.2% high Sep 2026 Official leaderboard
7 GPT-6 Astra 69% max Sep 2026 Official leaderboard
8 GPT-5.6 Terra 68.2% max Jul 2026 Official leaderboard
9 Claude Opus 4.5 67.9% 16k Nov 2025 Official leaderboard
10 GPT-6 Sol 67.8% max Sep 2026 Official leaderboard
11 GPT-5.5 66.8% xhigh Apr 2026 Official leaderboard
12 Claude Sonnet 5 66.5% max Jun 2026 Official leaderboard
13 Muse Wonder Full 66.5% Oct 2026 Lab model card
14 GLM-5.2 65.3% max Jun 2026 Official leaderboard
15 Gemini 3.7 Flash 64.7% high Aug 2026 Official leaderboard
16 Gemini 3.6 Flash 64% high Jul 2026 Official leaderboard
17 Kimi K3.1 Pro 63.5% max Oct 2026 Official leaderboard
18 Claude Opus 4.8 63.1% max May 2026 Official leaderboard
19 GPT-5.1 62.9% high Nov 2025 Official leaderboard
20 Gemini 3.5 Flash 62.7% high May 2026 Official leaderboard
21 Muse Spark 1.3 62.7% Oct 2026 Lab model card
22 Gemini 3.8 Flash 62.4% high Sep 2026 Official leaderboard
23 Muse Spark 1.1 61.9% Jul 2026 Lab model card
24 Grok 4.6 61.6% xhigh Aug 2026 Official leaderboard
25 Beam 1.0 61% Oct 2026 Lab model card
26 DeepSeek V4.1 Flash 1017 60.5% max Oct 2026 Official leaderboard
27 Claude Opus 4.1 60.1% 16k Aug 2025 Official leaderboard
28 GPT-5.2 60.1% xhigh Dec 2025 Official leaderboard
29 GPT-5 59.2% high Aug 2025 Official leaderboard
30 Kimi K3 58.5% max Jul 2026 Official leaderboard
31 o3 58.3% high Apr 2025 Official leaderboard
32 DeepSeek V4 Pro 0813 58% max Aug 2026 Official leaderboard
33 Qwen3.8-2.4T-A95B 57.8% xhigh Aug 2026 Official leaderboard
34 Qwen3.7-Max 56.8% May 2026 Official leaderboard
35 Claude Sonnet 4.5 56% 16k Sep 2025 Official leaderboard
36 Mistral Large 4 56% high Oct 2026 Official leaderboard
37 GPT-5 Pro 55.9% high Oct 2025 Official leaderboard
38 Muse Spark 55.7% Jun 2026 Lab model card
39 GLM-5.3 55.7% max Aug 2026 Official leaderboard
40 Inkling-Small 54.9% xhigh Jul 2026 Official leaderboard
41 o4-mini 54.5% high Apr 2025 Official leaderboard
42 GPT-5 mini 54.2% high Aug 2025 Official leaderboard
43 Grok 4.7 53.7% xhigh Sep 2026 Official leaderboard
44 Gemma 4 31B IT 53.6% Default Apr 2026 Official leaderboard
45 GLM-5 53.4% Feb 2026 Official leaderboard
46 Kimi K2 Thinking 52.4% Default Nov 2025 Official leaderboard
47 Nemotron 3.5 Super VL 52.4% Default Oct 2026 Official leaderboard
48 Grok 4 51.4% Default Jul 2025 Official leaderboard
49 Inkling 51.1% xhigh Jul 2026 Official leaderboard
50 Qwen 3.6 Plus 50.2% Mar 2026 Official leaderboard
51 Claude 3.5 Sonnet (October 2024) 49.9% Oct 2024 Official leaderboard
52 Grok 4.1 Fast 49.7% Default Nov 2025 Official leaderboard
53 Kimi K2.5 49.2% Default Jan 2026 Official leaderboard
54 Qwen3-235B-A22B-Thinking Jul 2025 49% Jul 2025 Official leaderboard
55 Qwen 3.8 27B 49% xhigh Aug 2026 Official leaderboard
56 Qwen3.5 397B-A17B 48.3% Feb 2026 Official leaderboard
57 o1 48.2% high Dec 2024 Official leaderboard
58 GPT-5.4 47.3% xhigh Mar 2026 Official leaderboard
59 Qwen3-Next-80B-A3B 47.2% thinking Sep 2025 Official leaderboard
60 Grok 4 Fast 46.8% Default Sep 2025 Official leaderboard
61 gpt-oss-120b 45.8% high Aug 2025 Official leaderboard
62 o3-mini 45.6% high Jan 2025 Official leaderboard
63 GPT-5 nano 45.4% high Aug 2025 Official leaderboard
64 Qwen3.6 27B 45.3% Default Apr 2026 Official leaderboard
65 DeepSeek-R1 May 2025 44.1% Default May 2025 Official leaderboard
66 GPT-4.5 43.3% Feb 2025 Official leaderboard
67 Claude 3.5 Sonnet 42.7% Jun 2024 Official leaderboard
68 Claude Sonnet 4 42% 16k May 2025 Official leaderboard
69 DeepSeek-V3.1 41.6% Default Aug 2025 Official leaderboard
70 Claude Haiku 4.5 41.6% 16k Oct 2025 Official leaderboard
71 Qwen3-235B-A22B 41.5% Apr 2025 Official leaderboard
72 Llama 4 Maverick 41.1% Apr 2025 Official leaderboard
73 Grok-3 mini 40.9% high Apr 2025 Official leaderboard
74 DeepSeek-V3.2 40% Default Dec 2025 Official leaderboard
75 GPT-4.1 mini 39% Apr 2025 Official leaderboard
76 DeepSeek-V4-Pro 38.8% max Apr 2026 Official leaderboard
77 GPT-4.1 38.7% Apr 2025 Official leaderboard
78 GPT-4o Nov 2024 36.2% Nov 2024 Official leaderboard
79 Nemotron 3 Nano 30B-A3B 36.2% Dec 2025 Official leaderboard
80 GPT-4.1 nano 35.2% Apr 2025 Official leaderboard
81 Mistral Medium 3.5 33.8% high Apr 2026 Official leaderboard
82 Qwen3-Next-80B-A3B-Instruct 33.5% Sep 2025 Official leaderboard
83 GPT-4o Aug 2024 33.4% Aug 2024 Official leaderboard
84 Grok 3 33% Apr 2025 Official leaderboard
85 Code World Model (CWM) 31.2% Oct 2025 Lab model card
86 Nemotron 3.5 Lightning 30B-A3B 26.5% Default Aug 2026 Official leaderboard
87 Claude 2 23.6% Jul 2023 Official leaderboard
88 Mistral Small 3.1 23.2% Mar 2025 Official leaderboard
89 Magistral Medium 1.0 22.9% Default Jun 2025 Official leaderboard
90 Mixtral 8x7B 15.1% Dec 2023 Official leaderboard
91 GPT-4o mini 12.9% Jul 2024 Official leaderboard

Labs report scores for 9 models in their cards. Switch the scores above to rank them together.

5 models have held number one on the official leaderboard since its release in August 2025 5 models have held number one with claims included since its release in August 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 and what labs claim in their cards.

OpenAI Anthropic Meta
Own leaderboard 5 leaders
Claude Opus 4.6 80.7%
Lab claims 3 leaders
GPT-5.6 Sol 68.4%

Sources: Official Human Pathogen Capabilities Test leaderboard and labs’ model cards

Scores over time

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 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 Opus 4.6 Claude Opus 4.5 GPT-5.1 o3 Claude 3.5 Sonnet (October 2024) Claude 3.5 Sonnet Claude 2 0% 20% 40% 60% 80% 100% 2024 2025 2026 2027 Near the ceiling (85%) Claude Opus 4.6 Claude Opus 4.5

Sources: Official Human Pathogen Capabilities Test leaderboard and labs’ model cards

Labs report it only in system cards

We read the headline results of 522 launch posts from 52 labs, and none of them list this benchmark. Some system cards do, below.

10 system cards report it, from 3 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 23, 2026 Muse Spark 1.3 system card 62.7 HPCT accuracy, Muse Spark 1.3 (API) max (Table 1, Figure 1) Bio and chem risk
Oct 12, 2026 Muse Wonder Full model card 66.5 SecureBio HPCT (CBRN Capability Evaluations table, section 4) Bio and chem risk
Oct 5, 2026 Beam 1.0 technical report 61 HPCT accuracy Bio and chem risk
Sep 3, 2026 GPT-6 Astra system card 65.9 SecureBio Human Pathogen Capabilities Test (HPCT), refusal-corrected accuracy (10.1.1.4) Bio and chem risk
Jul 9, 2026 GPT-5.6 Sol system card 68.4 SecureBio Human Pathogen Capabilities Test, strongest reported configuration (9.1.1.8) Bio and chem risk
Jul 9, 2026 Muse Spark 1.1 system card 61.9 HPCT, Muse Spark 1.1 Helpful (Table 1, Figure 1) Bio and chem risk
Jun 25, 2026 GPT-5.6 Sol system card 68.4 SecureBio Human Pathogen Capabilities Test, strongest reported configuration (9.1.1.8) Bio and chem risk
Jun 12, 2026 Muse Spark system card 55.7 HPCT (Table 1, Figure 1) Bio and chem risk
Oct 8, 2025 Code World Model (CWM) technical report 31.2 HPCT, multiple response (Table 13) Bio and chem risk
Aug 7, 2025 GPT-5 system card 53.7 SecureBio Human Pathogen Capabilities Test, mean of n=10 (Table 15) Bio and chem risk

The lab's number comes out ahead in 1 of 4 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. Every benchmark.

Model Lab says Board says Gap
GPT-5.6 Sol 68.4% SecureBio Human Pathogen Capabilities Test, strongest reported configuration (9.1.1.8) 76.3% Leaderboard, max −7.9 pts
GPT-5 53.7% SecureBio Human Pathogen Capabilities Test, mean of n=10 (Table 15) 59.2% Leaderboard, high −5.5 pts
GPT-6 Astra 65.9% SecureBio Human Pathogen Capabilities Test (HPCT), refusal-corrected accuracy (10.1.1.4) 69% Leaderboard, max −3.1 pts
Beam 1.0 61% HPCT accuracy 60.7% Leaderboard, max +0.3 pts

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 Anthropic
Human Pathogen Capabilities Test v2 (HPCT v2) 8 leaders
Claude Opus 5 80%
Human Pathogen Capabilities Test 5 leaders
Claude Opus 4.6 80.7%
Human Pathogen Capabilities Test v2 (HPCT v2) 3 leaders
Claude Opus 5 80%

Source: Official leaderboards

Replaced by Human Pathogen Capabilities Test v2 (HPCT v2)

More in Biology and chemistry

All 109 in Biology and chemistry
  • GeneBench Pro In use Agents analyze messy simulated genomics and biology datasets to estimate a decision-relevant quantity across 129 problems. Scored automatically against known answers. GPT-5.6 Sol 31.5%
  • SUPERChem In use Models solve 500 expert-written chemistry reasoning problems, offered in multimodal and text-only forms. Expert solution paths also allow a reasoning path score. Tencent Hy4 preview 66.4%
  • LifeSciBench In use Models complete 750 expert-written life science research tasks, often with attached figures, tables, or sequence files. Graded against expert rubrics as score and pass rate. GPT-Rosalind-6 61.3%
  • MedChemBench (OpenAI) In use Models answer chemistry questions from an OpenAI internal set, grouped with OpenAI's life science research evaluations. OpenAI has not published the task format. GPT-Rosalind-6 49.6%
  • ProtocolQA Open-Ended In use Models correct errors in published lab protocols using open-ended questions adapted from LAB-Bench ProtocolQA. Scored as pass@1. GPT-6 Cyber 47.2%
  • MolecularIQ In use Models answer questions about molecular graphs, such as counting, indexing and constrained generation, and a symbolic checker verifies each answer. gpt-oss-120b 47.5%