Benchmark Almanac

World-Class Bio

Claude Opus 5 68.8% leads since July 2026
Claude Opus 5 68.8% leads since July 2026
Only in system cards

Models answer expert-level biology questions in a SecureBio benchmark. Scored as accuracy.

96 text-only rubric-graded open-ended questions on rare biology known to a handful of experts. In development by SecureBio. Leading reasoning models score about 50%. Graded by an AI model with the rubric.

Made by
SecureBio
Kind
Genomics and life sciences
Level
Research frontier
Scored
Checked automatically
Answers
Short answer
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
86
Entries on its own leaderboard, 74 models we track
5
System cards report it, from 2 labs

Claude Opus 5 leads with 68.8%

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 5 leads with 68.8%, 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.

86 entries. Scroll sideways for the rest.

  1. 68.8 Claude Opus 5
  2. 66.7 Claude Opus 4.8
  3. 66.5 Claude Sonnet 5.5
  4. 64.4 GPT-5.6 Sol
  5. 62.6 GPT-6 Astra
  6. 61.1 Claude Opus 4.7
  7. 61.1 Gemini 4 Argon
  8. 60.5 GPT-6 Sol
  9. 59.8 GPT-5.6 Terra
  10. 59.7 GPT-5.5
  11. 58.0 Gemini 3.7 Flash
  12. 57.6 Grok 4.7
  13. 57.1 Gemini 3.8 Flash
  14. 56.7 Gemini 3.5 Flash
  15. 56.5 Claude Opus 4.6
  16. 56.5 GPT-5.5 pro Codex
  17. 56.3 GPT-5.2 Pro
  18. 55.4 Claude Sonnet 4.6
  19. 55.0 GPT-5.5 Rosalind
  20. 55.0 Kimi K3
  21. 54.6 GPT-5
  22. 54.5 Claude Opus 5.5
  23. 54.2 Gemini 3 Pro (Nov '25 Preview)
  24. 54.1 Kimi K2.6
  25. 53.8 o3
  26. 53.3 Gemini 3.1 Pro (Feb 26' Preview)
  27. 53.2 GPT-5.1
  28. 52.2 Gemini 3.6 Flash
  29. 51.5 Kimi K3.1 Pro
  30. 50.5 Qwen3.8-2.4T-A95B
  31. 50.3 Grok 4.6
  32. 50.2 Qwen3.7-Max
  33. 49.6 DeepSeek V4.1 Flash 1017
  34. 49.2 GPT-5 mini
  35. 49.1 Inkling
  36. 49.0 Claude Sonnet 5
  37. 48.8 Beam 1.0
  38. 47.9 Claude Opus 4.5
  39. 47.5 o4-mini
  40. 47.0 GLM-5.2
  41. 46.6 Muse Glimmer 30B
  42. 46.6 GLM-5.3
  43. 46.4 GPT-5.4
  44. 46.1 Gemini 3 Flash (Dec '25 Preview)
  45. 44.7 DeepSeek V4 Pro 0813
  46. 44.3 Mistral Large 4
  47. 44.2 Inkling-Small
  48. 43.6 GPT-5 Pro
  49. 43.6 Qwen 3.6 Plus
  50. 43.4 Nemotron 3.5 Super VL
  51. 43.2 Nemotron 3 Ultra 550B A55B
  52. 42.8 Claude Opus 4.1
  53. 40.7 gpt-oss-120b
  54. 40.3 Claude Opus 4
  55. 40.0 Qwen 3.8 27B
  56. 39.4 Gemini 2.5 Pro (snapshot unspecified)
  57. 38.9 Kimi K2 Thinking
  58. 38.4 GLM-5
  59. 37.9 Claude Sonnet 4.5
  60. 37.5 Kimi K2.5
  61. 37.3 DeepSeek-V4-Pro
  62. 34.7 Gemini 2.5 Flash (snapshot unspecified)
  63. 34.2 Qwen3.5 397B-A17B
  64. 33.7 Gemma 4 31B IT
  65. 33.6 GPT-5 nano
  66. 33.2 Qwen3.6 27B
  67. 33.0 Nemotron 3 Super 120B A12B
  68. 33.0 Mistral Medium 3.5
  69. 32.1 DeepSeek-V3.2
  70. 31.8 GPT-4.1
  71. 31.2 o1
  72. 30.9 Nemotron 3.5 Lightning 30B-A3B
  73. 30.6 o3-mini
  74. 28.6 DeepSeek-V3.1
  75. 27.0 Claude Sonnet 4
  76. 25.1 Nemotron 3 Nano 30B-A3B
  77. 24.4 GPT-4.1 mini
  78. 23.4 Claude Haiku 4.5
  79. 21.2 GPT-4o
  80. 17.2 GPT-4.1 nano
  81. 15.0 GPT-4o
  82. 13.1 GPT-4 Turbo (snapshot unspecified)
  83. 12.2 GPT-4 (snapshot unspecified)
  84. 9.50 GPT-4o mini
  85. 6.60 GPT-3.5 Turbo
  86. 4.20 GPT-4.5
  1. 1 Claude Opus 5 max 68.8%
  2. 2 Claude Opus 4.8 max 66.7%
  3. 3 Claude Sonnet 5.5 max 66.5%
  4. 4 GPT-5.6 Sol max 64.4%
  5. 5 GPT-6 Astra max 62.6%
  6. 6 Claude Opus 4.7 max 61.1%
  7. 7 Gemini 4 Argon high 61.1%
  8. 8 GPT-6 Sol max 60.5%
  9. 9 GPT-5.6 Terra max 59.8%
  10. 10 GPT-5.5 xhigh 59.7%
  11. 11 Gemini 3.7 Flash high 58%
  12. 12 Grok 4.7 xhigh 57.6%
  13. 13 Gemini 3.8 Flash high 57.1%
  14. 14 Gemini 3.5 Flash high 56.7%
  15. 15 Claude Opus 4.6 max 56.5%
  16. 16 GPT-5.5 pro Codex xhigh 56.5%
  17. 17 GPT-5.2 Pro xhigh 56.3%
  18. 18 Claude Sonnet 4.6 max 55.4%
  19. 19 GPT-5.5 Rosalind xhigh 55%
  20. 20 Kimi K3 max 55%
  21. 21 GPT-5 high 54.6%
  22. 22 Claude Opus 5.5 max 54.5%
  23. 23 Gemini 3 Pro (Nov '25 Preview) high 54.2%
  24. 24 Kimi K2.6 Default 54.1%
  25. 25 o3 high 53.8%
  26. 26 Gemini 3.1 Pro (Feb 26' Preview) high 53.3%
  27. 27 GPT-5.1 high 53.2%
  28. 28 Gemini 3.6 Flash high 52.2%
  29. 29 Kimi K3.1 Pro max 51.5%
  30. 30 Qwen3.8-2.4T-A95B xhigh 50.5%
  31. 31 Grok 4.6 xhigh 50.3%
  32. 32 Qwen3.7-Max 50.2%
  33. 33 DeepSeek V4.1 Flash 1017 max 49.6%
  34. 34 GPT-5 mini high 49.2%
  35. 35 Inkling xhigh 49.1%
  36. 36 Claude Sonnet 5 max 49%
  37. 37 Beam 1.0 max 48.8%
  38. 38 Claude Opus 4.5 16k 47.9%
  39. 39 o4-mini high 47.5%
  40. 40 GLM-5.2 max 47%
  41. 41 Muse Glimmer 30B xhigh 46.6%
  42. 42 GLM-5.3 max 46.6%
  43. 43 GPT-5.4 xhigh 46.4%
  44. 44 Gemini 3 Flash (Dec '25 Preview) high 46.1%
  45. 45 DeepSeek V4 Pro 0813 max 44.7%
  46. 46 Mistral Large 4 high 44.3%
  47. 47 Inkling-Small xhigh 44.2%
  48. 48 GPT-5 Pro high 43.6%
  49. 49 Qwen 3.6 Plus 43.6%
  50. 50 Nemotron 3.5 Super VL Default 43.4%
  51. 51 Nemotron 3 Ultra 550B A55B Default 43.2%
  52. 52 Claude Opus 4.1 16k 42.8%
  53. 53 gpt-oss-120b 40.7%
  54. 54 Claude Opus 4 16k 40.3%
  55. 55 Qwen 3.8 27B xhigh 40%
  56. 56 Gemini 2.5 Pro (snapshot unspecified) 16k 39.4%
  57. 57 Kimi K2 Thinking Default 38.9%
  58. 58 GLM-5 38.4%
  59. 59 Claude Sonnet 4.5 16k 37.9%
  60. 60 Kimi K2.5 Default 37.5%
  61. 61 DeepSeek-V4-Pro max 37.3%
  62. 62 Gemini 2.5 Flash (snapshot unspecified) 16k 34.7%
  63. 63 Qwen3.5 397B-A17B 34.2%
  64. 64 Gemma 4 31B IT Default 33.7%
  65. 65 GPT-5 nano high 33.6%
  66. 66 Qwen3.6 27B Default 33.2%
  67. 67 Nemotron 3 Super 120B A12B Default 33%
  68. 68 Mistral Medium 3.5 high 33%
  69. 69 DeepSeek-V3.2 Default 32.1%
  70. 70 GPT-4.1 31.8%
  71. 71 o1 high 31.2%
  72. 72 Nemotron 3.5 Lightning 30B-A3B Default 30.9%
  73. 73 o3-mini high 30.6%
  74. 74 DeepSeek-V3.1 Default 28.6%
  75. 75 Claude Sonnet 4 16k 27%
  76. 76 Nemotron 3 Nano 30B-A3B 25.1%
  77. 77 GPT-4.1 mini 24.4%
  78. 78 Claude Haiku 4.5 16k 23.4%
  79. 79 GPT-4o Nov 2024 21.2%
  80. 80 GPT-4.1 nano 17.2%
  81. 81 GPT-4o Aug 2024 15%
  82. 82 GPT-4 Turbo (snapshot unspecified) 13.1%
  83. 83 GPT-4 (snapshot unspecified) 12.2%
  84. 84 GPT-4o mini 9.50%
  85. 85 GPT-3.5 Turbo Jun 2023 6.60%
  86. 86 GPT-4.5 4.20%

Source: Official World-Class Bio leaderboard

74 models. Scroll sideways for the rest.

  1. 68.8 Claude Opus 5
  2. 68.3 GPT-5.6 Sol
  3. 66.7 Claude Opus 4.8
  4. 66.5 Claude Sonnet 5.5
  5. 62.6 GPT-6 Astra
  6. 61.1 Claude Opus 4.7
  7. 61.1 Gemini 4 Argon
  8. 60.5 GPT-6 Sol
  9. 59.8 GPT-5.6 Terra
  10. 59.7 GPT-5.5
  11. 58.0 Gemini 3.7 Flash
  12. 57.6 Grok 4.7
  13. 57.1 Gemini 3.8 Flash
  14. 56.7 Gemini 3.5 Flash
  15. 56.5 Claude Opus 4.6
  16. 56.3 GPT-5.2 Pro
  17. 55.4 Claude Sonnet 4.6
  18. 55.0 Kimi K3
  19. 54.6 GPT-5
  20. 54.5 Claude Opus 5.5
  21. 54.1 Kimi K2.6
  22. 53.8 o3
  23. 53.2 GPT-5.1
  24. 52.2 Gemini 3.6 Flash
  25. 51.5 Kimi K3.1 Pro
  26. 50.5 Qwen3.8-2.4T-A95B
  27. 50.3 Grok 4.6
  28. 50.2 Qwen3.7-Max
  29. 49.6 DeepSeek V4.1 Flash 1017
  30. 49.2 GPT-5 mini
  31. 49.1 Inkling
  32. 49.0 Claude Sonnet 5
  33. 48.8 Beam 1.0
  34. 47.9 Claude Opus 4.5
  35. 47.5 o4-mini
  36. 47.0 GLM-5.2
  37. 46.6 GLM-5.3
  38. 46.4 GPT-5.4
  39. 44.7 DeepSeek V4 Pro 0813
  40. 44.3 Mistral Large 4
  41. 44.2 Inkling-Small
  42. 43.6 GPT-5 Pro
  43. 43.6 Qwen 3.6 Plus
  44. 43.4 Nemotron 3.5 Super VL
  45. 42.8 Claude Opus 4.1
  46. 40.7 gpt-oss-120b
  47. 40.3 Claude Opus 4
  48. 40.0 Qwen 3.8 27B
  49. 38.9 Kimi K2 Thinking
  50. 38.4 GLM-5
  51. 37.9 Claude Sonnet 4.5
  52. 37.5 Kimi K2.5
  53. 37.3 DeepSeek-V4-Pro
  54. 34.2 Qwen3.5 397B-A17B
  55. 33.7 Gemma 4 31B IT
  56. 33.6 GPT-5 nano
  57. 33.2 Qwen3.6 27B
  58. 33.0 Mistral Medium 3.5
  59. 32.1 DeepSeek-V3.2
  60. 31.8 GPT-4.1
  61. 31.2 o1
  62. 30.9 Nemotron 3.5 Lightning 30B-A3B
  63. 30.6 o3-mini
  64. 28.6 DeepSeek-V3.1
  65. 27.0 Claude Sonnet 4
  66. 25.1 Nemotron 3 Nano 30B-A3B
  67. 24.4 GPT-4.1 mini
  68. 23.4 Claude Haiku 4.5
  69. 21.2 GPT-4o
  70. 17.2 GPT-4.1 nano
  71. 15.0 GPT-4o
  72. 9.50 GPT-4o mini
  73. 6.60 GPT-3.5 Turbo
  74. 4.20 GPT-4.5
  1. 1 Claude Opus 5 max 68.8%
  2. 2 GPT-5.6 Sol 68.3%
  3. 3 Claude Opus 4.8 max 66.7%
  4. 4 Claude Sonnet 5.5 max 66.5%
  5. 5 GPT-6 Astra max 62.6%
  6. 6 Claude Opus 4.7 max 61.1%
  7. 7 Gemini 4 Argon high 61.1%
  8. 8 GPT-6 Sol max 60.5%
  9. 9 GPT-5.6 Terra max 59.8%
  10. 10 GPT-5.5 xhigh 59.7%
  11. 11 Gemini 3.7 Flash high 58%
  12. 12 Grok 4.7 xhigh 57.6%
  13. 13 Gemini 3.8 Flash high 57.1%
  14. 14 Gemini 3.5 Flash high 56.7%
  15. 15 Claude Opus 4.6 max 56.5%
  16. 16 GPT-5.2 Pro xhigh 56.3%
  17. 17 Claude Sonnet 4.6 max 55.4%
  18. 18 Kimi K3 max 55%
  19. 19 GPT-5 high 54.6%
  20. 20 Claude Opus 5.5 max 54.5%
  21. 21 Kimi K2.6 Default 54.1%
  22. 22 o3 high 53.8%
  23. 23 GPT-5.1 high 53.2%
  24. 24 Gemini 3.6 Flash high 52.2%
  25. 25 Kimi K3.1 Pro max 51.5%
  26. 26 Qwen3.8-2.4T-A95B xhigh 50.5%
  27. 27 Grok 4.6 xhigh 50.3%
  28. 28 Qwen3.7-Max 50.2%
  29. 29 DeepSeek V4.1 Flash 1017 max 49.6%
  30. 30 GPT-5 mini high 49.2%
  31. 31 Inkling xhigh 49.1%
  32. 32 Claude Sonnet 5 max 49%
  33. 33 Beam 1.0 max 48.8%
  34. 34 Claude Opus 4.5 16k 47.9%
  35. 35 o4-mini high 47.5%
  36. 36 GLM-5.2 max 47%
  37. 37 GLM-5.3 max 46.6%
  38. 38 GPT-5.4 xhigh 46.4%
  39. 39 DeepSeek V4 Pro 0813 max 44.7%
  40. 40 Mistral Large 4 high 44.3%
  41. 41 Inkling-Small xhigh 44.2%
  42. 42 GPT-5 Pro high 43.6%
  43. 43 Qwen 3.6 Plus 43.6%
  44. 44 Nemotron 3.5 Super VL Default 43.4%
  45. 45 Claude Opus 4.1 16k 42.8%
  46. 46 gpt-oss-120b 40.7%
  47. 47 Claude Opus 4 16k 40.3%
  48. 48 Qwen 3.8 27B xhigh 40%
  49. 49 Kimi K2 Thinking Default 38.9%
  50. 50 GLM-5 38.4%
  51. 51 Claude Sonnet 4.5 16k 37.9%
  52. 52 Kimi K2.5 Default 37.5%
  53. 53 DeepSeek-V4-Pro max 37.3%
  54. 54 Qwen3.5 397B-A17B 34.2%
  55. 55 Gemma 4 31B IT Default 33.7%
  56. 56 GPT-5 nano high 33.6%
  57. 57 Qwen3.6 27B Default 33.2%
  58. 58 Mistral Medium 3.5 high 33%
  59. 59 DeepSeek-V3.2 Default 32.1%
  60. 60 GPT-4.1 31.8%
  61. 61 o1 high 31.2%
  62. 62 Nemotron 3.5 Lightning 30B-A3B Default 30.9%
  63. 63 o3-mini high 30.6%
  64. 64 DeepSeek-V3.1 Default 28.6%
  65. 65 Claude Sonnet 4 16k 27%
  66. 66 Nemotron 3 Nano 30B-A3B 25.1%
  67. 67 GPT-4.1 mini 24.4%
  68. 68 Claude Haiku 4.5 16k 23.4%
  69. 69 GPT-4o Nov 2024 21.2%
  70. 70 GPT-4.1 nano 17.2%
  71. 71 GPT-4o Aug 2024 15%
  72. 72 GPT-4o mini 9.50%
  73. 73 GPT-3.5 Turbo Jun 2023 6.60%
  74. 74 GPT-4.5 4.20%

Sources: Official World-Class Bio leaderboard and labs’ model cards

#EntryScoreSettingDateSource
1 Claude Opus 5 68.8% max Jul 2026 Official leaderboard
2 Claude Opus 4.8 66.7% max May 2026 Official leaderboard
3 Claude Sonnet 5.5 66.5% max Sep 2026 Official leaderboard
4 GPT-5.6 Sol 64.4% max Jul 2026 Official leaderboard
5 GPT-6 Astra 62.6% max Sep 2026 Official leaderboard
6 Claude Opus 4.7 61.1% max Apr 2026 Official leaderboard
7 Gemini 4 Argon 61.1% high Sep 2026 Official leaderboard
8 GPT-6 Sol 60.5% max Sep 2026 Official leaderboard
9 GPT-5.6 Terra 59.8% max Jul 2026 Official leaderboard
10 GPT-5.5 59.7% xhigh Apr 2026 Official leaderboard
11 Gemini 3.7 Flash 58% high Aug 2026 Official leaderboard
12 Grok 4.7 57.6% xhigh Sep 2026 Official leaderboard
13 Gemini 3.8 Flash 57.1% high Sep 2026 Official leaderboard
14 Gemini 3.5 Flash 56.7% high May 2026 Official leaderboard
15 Claude Opus 4.6 56.5% max Feb 2026 Official leaderboard
16 GPT-5.5 pro Codex 56.5% xhigh Apr 2026 Official leaderboard
17 GPT-5.2 Pro 56.3% xhigh Dec 2025 Official leaderboard
18 Claude Sonnet 4.6 55.4% max Feb 2026 Official leaderboard
19 GPT-5.5 Rosalind 55% xhigh Jun 2026 Official leaderboard
20 Kimi K3 55% max Jul 2026 Official leaderboard
21 GPT-5 54.6% high Aug 2025 Official leaderboard
22 Claude Opus 5.5 54.5% max Sep 2026 Official leaderboard
23 Gemini 3 Pro (Nov '25 Preview) 54.2% high Nov 2025 Official leaderboard
24 Kimi K2.6 54.1% Default Apr 2026 Official leaderboard
25 o3 53.8% high Apr 2025 Official leaderboard
26 Gemini 3.1 Pro (Feb 26' Preview) 53.3% high Feb 2026 Official leaderboard
27 GPT-5.1 53.2% high Nov 2025 Official leaderboard
28 Gemini 3.6 Flash 52.2% high Jul 2026 Official leaderboard
29 Kimi K3.1 Pro 51.5% max Oct 2026 Official leaderboard
30 Qwen3.8-2.4T-A95B 50.5% xhigh Aug 2026 Official leaderboard
31 Grok 4.6 50.3% xhigh Aug 2026 Official leaderboard
32 Qwen3.7-Max 50.2% May 2026 Official leaderboard
33 DeepSeek V4.1 Flash 1017 49.6% max Oct 2026 Official leaderboard
34 GPT-5 mini 49.2% high Aug 2025 Official leaderboard
35 Inkling 49.1% xhigh Jul 2026 Official leaderboard
36 Claude Sonnet 5 49% max Jun 2026 Official leaderboard
37 Beam 1.0 48.8% max Oct 2026 Official leaderboard
38 Claude Opus 4.5 47.9% 16k Nov 2025 Official leaderboard
39 o4-mini 47.5% high Apr 2025 Official leaderboard
40 GLM-5.2 47% max Jun 2026 Official leaderboard
41 Muse Glimmer 30B 46.6% xhigh Aug 2026 Official leaderboard
42 GLM-5.3 46.6% max Aug 2026 Official leaderboard
43 GPT-5.4 46.4% xhigh Mar 2026 Official leaderboard
44 Gemini 3 Flash (Dec '25 Preview) 46.1% high Dec 2025 Official leaderboard
45 DeepSeek V4 Pro 0813 44.7% max Aug 2026 Official leaderboard
46 Mistral Large 4 44.3% high Oct 2026 Official leaderboard
47 Inkling-Small 44.2% xhigh Jul 2026 Official leaderboard
48 GPT-5 Pro 43.6% high Oct 2025 Official leaderboard
49 Qwen 3.6 Plus 43.6% Mar 2026 Official leaderboard
50 Nemotron 3.5 Super VL 43.4% Default Oct 2026 Official leaderboard
51 Nemotron 3 Ultra 550B A55B 43.2% Default Jun 2026 Official leaderboard
52 Claude Opus 4.1 42.8% 16k Aug 2025 Official leaderboard
53 gpt-oss-120b 40.7% Aug 2025 Official leaderboard
54 Claude Opus 4 40.3% 16k May 2025 Official leaderboard
55 Qwen 3.8 27B 40% xhigh Aug 2026 Official leaderboard
56 Gemini 2.5 Pro (snapshot unspecified) 39.4% 16k Jun 2025 Official leaderboard
57 Kimi K2 Thinking 38.9% Default Nov 2025 Official leaderboard
58 GLM-5 38.4% Feb 2026 Official leaderboard
59 Claude Sonnet 4.5 37.9% 16k Sep 2025 Official leaderboard
60 Kimi K2.5 37.5% Default Jan 2026 Official leaderboard
61 DeepSeek-V4-Pro 37.3% max Apr 2026 Official leaderboard
62 Gemini 2.5 Flash (snapshot unspecified) 34.7% 16k Jun 2025 Official leaderboard
63 Qwen3.5 397B-A17B 34.2% Feb 2026 Official leaderboard
64 Gemma 4 31B IT 33.7% Default Apr 2026 Official leaderboard
65 GPT-5 nano 33.6% high Aug 2025 Official leaderboard
66 Qwen3.6 27B 33.2% Default Apr 2026 Official leaderboard
67 Nemotron 3 Super 120B A12B 33% Default Mar 2026 Official leaderboard
68 Mistral Medium 3.5 33% high Apr 2026 Official leaderboard
69 DeepSeek-V3.2 32.1% Default Dec 2025 Official leaderboard
70 GPT-4.1 31.8% Apr 2025 Official leaderboard
71 o1 31.2% high Dec 2024 Official leaderboard
72 Nemotron 3.5 Lightning 30B-A3B 30.9% Default Aug 2026 Official leaderboard
73 o3-mini 30.6% high Jan 2025 Official leaderboard
74 DeepSeek-V3.1 28.6% Default Aug 2025 Official leaderboard
75 Claude Sonnet 4 27% 16k May 2025 Official leaderboard
76 Nemotron 3 Nano 30B-A3B 25.1% Dec 2025 Official leaderboard
77 GPT-4.1 mini 24.4% Apr 2025 Official leaderboard
78 Claude Haiku 4.5 23.4% 16k Oct 2025 Official leaderboard
79 GPT-4o Nov 2024 21.2% Nov 2024 Official leaderboard
80 GPT-4.1 nano 17.2% Apr 2025 Official leaderboard
81 GPT-4o Aug 2024 15% Aug 2024 Official leaderboard
82 GPT-4 Turbo (snapshot unspecified) 13.1% Nov 2023 Official leaderboard
83 GPT-4 (snapshot unspecified) 12.2% Mar 2023 Official leaderboard
84 GPT-4o mini 9.50% Jul 2024 Official leaderboard
85 GPT-3.5 Turbo Jun 2023 6.60% Jun 2023 Official leaderboard
86 GPT-4.5 4.20% Feb 2025 Official leaderboard
#ModelScoreSettingDateSource
1 Claude Opus 5 68.8% max Jul 2026 Official leaderboard
2 GPT-5.6 Sol 68.3% Jun 2026 Lab model card
3 Claude Opus 4.8 66.7% max May 2026 Official leaderboard
4 Claude Sonnet 5.5 66.5% max Sep 2026 Official leaderboard
5 GPT-6 Astra 62.6% max Sep 2026 Official leaderboard
6 Claude Opus 4.7 61.1% max Apr 2026 Official leaderboard
7 Gemini 4 Argon 61.1% high Sep 2026 Official leaderboard
8 GPT-6 Sol 60.5% max Sep 2026 Official leaderboard
9 GPT-5.6 Terra 59.8% max Jul 2026 Official leaderboard
10 GPT-5.5 59.7% xhigh Apr 2026 Official leaderboard
11 Gemini 3.7 Flash 58% high Aug 2026 Official leaderboard
12 Grok 4.7 57.6% xhigh Sep 2026 Official leaderboard
13 Gemini 3.8 Flash 57.1% high Sep 2026 Official leaderboard
14 Gemini 3.5 Flash 56.7% high May 2026 Official leaderboard
15 Claude Opus 4.6 56.5% max Feb 2026 Official leaderboard
16 GPT-5.2 Pro 56.3% xhigh Dec 2025 Official leaderboard
17 Claude Sonnet 4.6 55.4% max Feb 2026 Official leaderboard
18 Kimi K3 55% max Jul 2026 Official leaderboard
19 GPT-5 54.6% high Aug 2025 Official leaderboard
20 Claude Opus 5.5 54.5% max Sep 2026 Official leaderboard
21 Kimi K2.6 54.1% Default Apr 2026 Official leaderboard
22 o3 53.8% high Apr 2025 Official leaderboard
23 GPT-5.1 53.2% high Nov 2025 Official leaderboard
24 Gemini 3.6 Flash 52.2% high Jul 2026 Official leaderboard
25 Kimi K3.1 Pro 51.5% max Oct 2026 Official leaderboard
26 Qwen3.8-2.4T-A95B 50.5% xhigh Aug 2026 Official leaderboard
27 Grok 4.6 50.3% xhigh Aug 2026 Official leaderboard
28 Qwen3.7-Max 50.2% May 2026 Official leaderboard
29 DeepSeek V4.1 Flash 1017 49.6% max Oct 2026 Official leaderboard
30 GPT-5 mini 49.2% high Aug 2025 Official leaderboard
31 Inkling 49.1% xhigh Jul 2026 Official leaderboard
32 Claude Sonnet 5 49% max Jun 2026 Official leaderboard
33 Beam 1.0 48.8% max Oct 2026 Official leaderboard
34 Claude Opus 4.5 47.9% 16k Nov 2025 Official leaderboard
35 o4-mini 47.5% high Apr 2025 Official leaderboard
36 GLM-5.2 47% max Jun 2026 Official leaderboard
37 GLM-5.3 46.6% max Aug 2026 Official leaderboard
38 GPT-5.4 46.4% xhigh Mar 2026 Official leaderboard
39 DeepSeek V4 Pro 0813 44.7% max Aug 2026 Official leaderboard
40 Mistral Large 4 44.3% high Oct 2026 Official leaderboard
41 Inkling-Small 44.2% xhigh Jul 2026 Official leaderboard
42 GPT-5 Pro 43.6% high Oct 2025 Official leaderboard
43 Qwen 3.6 Plus 43.6% Mar 2026 Official leaderboard
44 Nemotron 3.5 Super VL 43.4% Default Oct 2026 Official leaderboard
45 Claude Opus 4.1 42.8% 16k Aug 2025 Official leaderboard
46 gpt-oss-120b 40.7% Aug 2025 Official leaderboard
47 Claude Opus 4 40.3% 16k May 2025 Official leaderboard
48 Qwen 3.8 27B 40% xhigh Aug 2026 Official leaderboard
49 Kimi K2 Thinking 38.9% Default Nov 2025 Official leaderboard
50 GLM-5 38.4% Feb 2026 Official leaderboard
51 Claude Sonnet 4.5 37.9% 16k Sep 2025 Official leaderboard
52 Kimi K2.5 37.5% Default Jan 2026 Official leaderboard
53 DeepSeek-V4-Pro 37.3% max Apr 2026 Official leaderboard
54 Qwen3.5 397B-A17B 34.2% Feb 2026 Official leaderboard
55 Gemma 4 31B IT 33.7% Default Apr 2026 Official leaderboard
56 GPT-5 nano 33.6% high Aug 2025 Official leaderboard
57 Qwen3.6 27B 33.2% Default Apr 2026 Official leaderboard
58 Mistral Medium 3.5 33% high Apr 2026 Official leaderboard
59 DeepSeek-V3.2 32.1% Default Dec 2025 Official leaderboard
60 GPT-4.1 31.8% Apr 2025 Official leaderboard
61 o1 31.2% high Dec 2024 Official leaderboard
62 Nemotron 3.5 Lightning 30B-A3B 30.9% Default Aug 2026 Official leaderboard
63 o3-mini 30.6% high Jan 2025 Official leaderboard
64 DeepSeek-V3.1 28.6% Default Aug 2025 Official leaderboard
65 Claude Sonnet 4 27% 16k May 2025 Official leaderboard
66 Nemotron 3 Nano 30B-A3B 25.1% Dec 2025 Official leaderboard
67 GPT-4.1 mini 24.4% Apr 2025 Official leaderboard
68 Claude Haiku 4.5 23.4% 16k Oct 2025 Official leaderboard
69 GPT-4o Nov 2024 21.2% Nov 2024 Official leaderboard
70 GPT-4.1 nano 17.2% Apr 2025 Official leaderboard
71 GPT-4o Aug 2024 15% Aug 2024 Official leaderboard
72 GPT-4o mini 9.50% Jul 2024 Official leaderboard
73 GPT-3.5 Turbo Jun 2023 6.60% Jun 2023 Official leaderboard
74 GPT-4.5 4.20% Feb 2025 Official leaderboard

Labs report scores for 4 models in their cards. Switch the scores above to rank them together.

13 models have held number one on the official leaderboard since 2023 14 models have held number one with claims included since 2023

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
Own leaderboard 13 leaders
Claude Opus 5 68.8%
Lab claims 2 leaders
GPT-5.6 Sol 68.3%

Sources: Official World-Class Bio 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% 2023 2024 2025 2026 2027 Near the ceiling (85%) Claude Opus 5 Claude Opus 4.7 Claude Opus 4.6 o3 GPT-4.1 GPT-4o GPT-4o 0% 20% 40% 60% 80% 100% 2023 2024 2025 2026 2027 Near the ceiling (85%) Claude Opus 5 Claude Opus 4.7

Sources: Official World-Class Bio 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.

5 system cards report it, from 2 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 5, 2026 Beam 1.0 technical report 37.9 World-Class Biology accuracy Bio and chem risk
Sep 3, 2026 GPT-6 Astra system card 59.2 SecureBio World Class Bio (WCB), refusal-corrected accuracy, web search disabled, corrected 2026-10-30 (10.1.1.4, change log) Bio and chem risk
Jul 9, 2026 GPT-5.6 Sol system card 68.3 SecureBio World-Class Bio, strongest reported configuration (9.1.1.8) Bio and chem risk
Jun 25, 2026 GPT-5.6 Sol system card 68.3 SecureBio World-Class Bio, strongest reported configuration (9.1.1.8) Bio and chem risk
Aug 7, 2025 GPT-5 system card 47.6 SecureBio World Class Biology, 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
Beam 1.0 37.9% World-Class Biology accuracy 48.8% Leaderboard, max −11 pts
GPT-5 47.6% SecureBio World Class Biology, mean of n=10 (Table 15) 54.6% Leaderboard, high −7.0 pts
GPT-5.6 Sol 68.3% SecureBio World-Class Bio, strongest reported configuration (9.1.1.8) 64.4% Leaderboard, max +3.9 pts
GPT-6 Astra 59.2% SecureBio World Class Bio (WCB), refusal-corrected accuracy, web search disabled, corrected 2026-10-30 (10.1.1.4, change log) 62.6% Leaderboard, max −3.4 pts

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%