Benchmark Almanac

BioMysteryBench

Claude Fable 5.2 83.7% leads since November 2026
Claude Fable 5.1 90.3% leads since September 2026
In use

Bioinformatics problems on anonymized real datasets that require real analysis with tools. Rubric-graded, with human-solvable and human-difficult splits.

Version v11 (July 6th, 2026) removed 9 problems with wrong or underivable keys and edited 24, leaving 90 (73 human-solvable, 17 human-hard). Cards report the two splits apart. A 5-problem preview is open and the full set is gated.

Made by
Anthropic
Released
April 2026
Kind
Lab and data work
Level
Expert
Scored
Checked automatically
Answers
Short answer
Tasks
All public
Strongest model tested
Claude Fable 5.2 #1 on the capability index
Last in a launch post
November 2026
Last model added
November 2026
Scores refreshed
November 12th, 2026
31
Entries on Vals AI’s leaderboard, 30 models we track
7
Launch posts list it, from 5 labs
11
System cards report it, from 2 labs

Claude Fable 5.2 leads with 83.7%

Best score per entry from Vals AI’s 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 Fable 5.1 leads with 90.3%, 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.

31 entries. Scroll sideways for the rest.

  1. 83.7 Claude Fable 5.2
  2. 81.1 Claude Sonnet 5.5
  3. 79.6 GPT-6.1 Sol
  4. 79.3 Claude Opus 5
  5. 79.3 GPT-6 Astra
  6. 79.3 Claude Opus 5.5
  7. 76.3 Gemini 4 Argon
  8. 75.6 Kimi K3.1 Pro
  9. 74.8 GPT-6 Sol
  10. 73.3 Muse Wonder Full
  11. 72.6 Kimi K3
  12. 72.2 Grok 4.6
  13. 71.1 GPT-5.6 Sol
  14. 70.4 Claude Haiku 5.5
  15. 70.0 Inkling 2 Small
  16. 69.3 Tencent Hy4 preview
  17. 69.3 Grok 4.7
  18. 69.3 MiMo-V2.6 Flash
  19. 67.8 DeepSeek V4.1 Flash
  20. 67.4 Step 5 Preview
  21. 67.0 Wonder
  22. 67.0 Ember-1
  23. 67.0 Mistral Large 4 Preview
  24. 64.8 Muse Spark 1.2
  25. 64.4 DeepSeek V4 Flash 0731
  26. 62.2 Gemini 3.8 Flash
  27. 61.5 GPT-5.6 Luna
  28. 61.5 GPT-6 Luna
  29. 60.0 Gemini 3.1 Pro
  30. 58.5 Gemini 3.6 Flash
  31. 42.6 Beam 1.0
  1. 1 Claude Fable 5.2 Vals AI harness, v11 full set 83.7%
  2. 2 Claude Sonnet 5.5 Vals AI harness, v11 full set 81.1%
  3. 3 GPT-6.1 Sol Vals AI harness, v11 full set 79.6%
  4. 4 Claude Opus 5 Vals AI harness, v11 full set 79.3%
  5. 5 GPT-6 Astra Vals AI harness, v11 full set 79.3%
  6. 6 Claude Opus 5.5 Vals AI harness, v11 full set 79.3%
  7. 7 Gemini 4 Argon Vals AI harness, v11 full set 76.3%
  8. 8 Kimi K3.1 Pro Vals AI harness, v11 full set 75.6%
  9. 9 GPT-6 Sol Vals AI harness, v11 full set 74.8%
  10. 10 Muse Wonder Full Vals AI harness, v11 full set 73.3%
  11. 11 Kimi K3 Vals AI harness, v11 full set 72.6%
  12. 12 Grok 4.6 Vals AI harness, v11 full set 72.2%
  13. 13 GPT-5.6 Sol Vals AI harness, v11 full set 71.1%
  14. 14 Claude Haiku 5.5 Vals AI harness, v11 full set 70.4%
  15. 15 Inkling 2 Small Vals AI harness, v11 full set 70%
  16. 16 Tencent Hy4 preview Vals AI harness, v11 full set 69.3%
  17. 17 Grok 4.7 Vals AI harness, v11 full set 69.3%
  18. 18 MiMo-V2.6 Flash Vals AI harness, v11 full set 69.3%
  19. 19 DeepSeek V4.1 Flash Vals AI harness, v11 full set 67.8%
  20. 20 Step 5 Preview Vals AI harness, v11 full set 67.4%
  21. 21 Wonder Vals AI harness, v11 full set 67%
  22. 22 Ember-1 Vals AI harness, v11 full set 67%
  23. 23 Mistral Large 4 Preview Vals AI harness, v11 full set 67%
  24. 24 Muse Spark 1.2 Vals AI harness, v11 full set 64.8%
  25. 25 DeepSeek V4 Flash 0731 Vals AI harness, v11 full set 64.4%
  26. 26 Gemini 3.8 Flash Vals AI harness, v11 full set 62.2%
  27. 27 GPT-5.6 Luna Vals AI harness, v11 full set 61.5%
  28. 28 GPT-6 Luna Vals AI harness, v11 full set 61.5%
  29. 29 Gemini 3.1 Pro Vals AI harness, v11 full set 60%
  30. 30 Gemini 3.6 Flash Vals AI harness, v11 full set 58.5%
  31. 31 Beam 1.0 Vals AI harness, v11 full set 42.6%

Source: Vals AI’s leaderboard

35 models. Scroll sideways for the rest.

  1. 90.3 Claude Fable 5.1
  2. 90.1 Claude Opus 5
  3. 88.8 Gemini 3.8 Flash
  4. 87.1 Gemini 3.7 Flash
  5. 83.9 Claude Fable 5
  6. 83.7 Claude Fable 5.2
  7. 81.1 Claude Sonnet 5.5
  8. 80.6 Kimi K3.1 Pro
  9. 79.6 GPT-6.1 Sol
  10. 79.3 GPT-6 Astra
  11. 79.3 Claude Opus 5.5
  12. 76.3 Gemini 4 Argon
  13. 74.8 GPT-6 Sol
  14. 73.3 Muse Wonder Full
  15. 72.6 Kimi K3
  16. 72.2 Grok 4.6
  17. 71.3 Tencent Hy4 preview
  18. 71.1 GPT-5.6 Sol
  19. 70.4 Claude Haiku 5.5
  20. 70.0 Inkling 2 Small
  21. 69.3 Grok 4.7
  22. 69.3 MiMo-V2.6 Flash
  23. 67.8 DeepSeek V4.1 Flash
  24. 67.4 Step 5 Preview
  25. 67.0 Ember-1
  26. 67.0 Mistral Large 4 Preview
  27. 64.8 Muse Spark 1.2
  28. 64.4 DeepSeek V4 Flash 0731
  29. 61.5 Claude Opus 4.6
  30. 61.5 GPT-5.6 Luna
  31. 61.5 GPT-6 Luna
  32. 60.0 Gemini 3.1 Pro
  33. 58.5 Gemini 3.6 Flash
  34. 50.4 Claude Sonnet 4.6
  35. 42.6 Beam 1.0
  1. 1 Claude Fable 5.1 90.3%
  2. 2 Claude Opus 5 90.1%
  3. 3 Gemini 3.8 Flash 88.8%
  4. 4 Gemini 3.7 Flash 87.1%
  5. 5 Claude Fable 5 83.9%
  6. 6 Claude Fable 5.2 Vals AI harness, v11 full set 83.7%
  7. 7 Claude Sonnet 5.5 Vals AI harness, v11 full set 81.1%
  8. 8 Kimi K3.1 Pro 80.6%
  9. 9 GPT-6.1 Sol Vals AI harness, v11 full set 79.6%
  10. 10 GPT-6 Astra Vals AI harness, v11 full set 79.3%
  11. 11 Claude Opus 5.5 Vals AI harness, v11 full set 79.3%
  12. 12 Gemini 4 Argon Vals AI harness, v11 full set 76.3%
  13. 13 GPT-6 Sol Vals AI harness, v11 full set 74.8%
  14. 14 Muse Wonder Full Vals AI harness, v11 full set 73.3%
  15. 15 Kimi K3 Vals AI harness, v11 full set 72.6%
  16. 16 Grok 4.6 Vals AI harness, v11 full set 72.2%
  17. 17 Tencent Hy4 preview 71.3%
  18. 18 GPT-5.6 Sol Vals AI harness, v11 full set 71.1%
  19. 19 Claude Haiku 5.5 Vals AI harness, v11 full set 70.4%
  20. 20 Inkling 2 Small Vals AI harness, v11 full set 70%
  21. 21 Grok 4.7 Vals AI harness, v11 full set 69.3%
  22. 22 MiMo-V2.6 Flash Vals AI harness, v11 full set 69.3%
  23. 23 DeepSeek V4.1 Flash Vals AI harness, v11 full set 67.8%
  24. 24 Step 5 Preview Vals AI harness, v11 full set 67.4%
  25. 25 Ember-1 Vals AI harness, v11 full set 67%
  26. 26 Mistral Large 4 Preview Vals AI harness, v11 full set 67%
  27. 27 Muse Spark 1.2 Vals AI harness, v11 full set 64.8%
  28. 28 DeepSeek V4 Flash 0731 Vals AI harness, v11 full set 64.4%
  29. 29 Claude Opus 4.6 61.5%
  30. 30 GPT-5.6 Luna Vals AI harness, v11 full set 61.5%
  31. 31 GPT-6 Luna Vals AI harness, v11 full set 61.5%
  32. 32 Gemini 3.1 Pro Vals AI harness, v11 full set 60%
  33. 33 Gemini 3.6 Flash Vals AI harness, v11 full set 58.5%
  34. 34 Claude Sonnet 4.6 50.4%
  35. 35 Beam 1.0 Vals AI harness, v11 full set 42.6%

Sources: Labs’ model cards and Vals AI’s leaderboard

#EntryScoreSettingDateSource
1 Claude Fable 5.2 83.7% Vals AI harness, v11 full set Nov 2026 Official leaderboard
2 Claude Sonnet 5.5 81.1% Vals AI harness, v11 full set Sep 2026 Official leaderboard
3 GPT-6.1 Sol 79.6% Vals AI harness, v11 full set Sep 2026 Official leaderboard
4 Claude Opus 5 79.3% Vals AI harness, v11 full set Jul 2026 Official leaderboard
5 GPT-6 Astra 79.3% Vals AI harness, v11 full set Sep 2026 Official leaderboard
6 Claude Opus 5.5 79.3% Vals AI harness, v11 full set Sep 2026 Official leaderboard
7 Gemini 4 Argon 76.3% Vals AI harness, v11 full set Sep 2026 Official leaderboard
8 Kimi K3.1 Pro 75.6% Vals AI harness, v11 full set Oct 2026 Official leaderboard
9 GPT-6 Sol 74.8% Vals AI harness, v11 full set Sep 2026 Official leaderboard
10 Muse Wonder Full 73.3% Vals AI harness, v11 full set Oct 2026 Official leaderboard
11 Kimi K3 72.6% Vals AI harness, v11 full set Jul 2026 Official leaderboard
12 Grok 4.6 72.2% Vals AI harness, v11 full set Aug 2026 Official leaderboard
13 GPT-5.6 Sol 71.1% Vals AI harness, v11 full set Jul 2026 Official leaderboard
14 Claude Haiku 5.5 70.4% Vals AI harness, v11 full set Oct 2026 Official leaderboard
15 Inkling 2 Small 70% Vals AI harness, v11 full set Nov 2026 Official leaderboard
16 Tencent Hy4 preview 69.3% Vals AI harness, v11 full set Aug 2026 Official leaderboard
17 Grok 4.7 69.3% Vals AI harness, v11 full set Sep 2026 Official leaderboard
18 MiMo-V2.6 Flash 69.3% Vals AI harness, v11 full set Sep 2026 Official leaderboard
19 DeepSeek V4.1 Flash 67.8% Vals AI harness, v11 full set Sep 2026 Official leaderboard
20 Step 5 Preview 67.4% Vals AI harness, v11 full set Sep 2026 Official leaderboard
21 Wonder 67% Vals AI harness, v11 full set Official leaderboard
22 Ember-1 67% Vals AI harness, v11 full set Sep 2026 Official leaderboard
23 Mistral Large 4 Preview 67% Vals AI harness, v11 full set Oct 2026 Official leaderboard
24 Muse Spark 1.2 64.8% Vals AI harness, v11 full set Aug 2026 Official leaderboard
25 DeepSeek V4 Flash 0731 64.4% Vals AI harness, v11 full set Jul 2026 Official leaderboard
26 Gemini 3.8 Flash 62.2% Vals AI harness, v11 full set Sep 2026 Official leaderboard
27 GPT-5.6 Luna 61.5% Vals AI harness, v11 full set Jul 2026 Official leaderboard
28 GPT-6 Luna 61.5% Vals AI harness, v11 full set Sep 2026 Official leaderboard
29 Gemini 3.1 Pro 60% Vals AI harness, v11 full set Feb 2026 Official leaderboard
30 Gemini 3.6 Flash 58.5% Vals AI harness, v11 full set Jul 2026 Official leaderboard
31 Beam 1.0 42.6% Vals AI harness, v11 full set Oct 2026 Official leaderboard
#ModelScoreSettingDateSource
1 Claude Fable 5.1 90.3% Sep 2026 Lab model card
2 Claude Opus 5 90.1% Jul 2026 Lab model card
3 Gemini 3.8 Flash 88.8% Sep 2026 Lab model card
4 Gemini 3.7 Flash 87.1% Aug 2026 Lab model card
5 Claude Fable 5 83.9% Jun 2026 Lab model card
6 Claude Fable 5.2 83.7% Vals AI harness, v11 full set Nov 2026 Official leaderboard
7 Claude Sonnet 5.5 81.1% Vals AI harness, v11 full set Sep 2026 Official leaderboard
8 Kimi K3.1 Pro 80.6% Oct 2026 Lab model card
9 GPT-6.1 Sol 79.6% Vals AI harness, v11 full set Sep 2026 Official leaderboard
10 GPT-6 Astra 79.3% Vals AI harness, v11 full set Sep 2026 Official leaderboard
11 Claude Opus 5.5 79.3% Vals AI harness, v11 full set Sep 2026 Official leaderboard
12 Gemini 4 Argon 76.3% Vals AI harness, v11 full set Sep 2026 Official leaderboard
13 GPT-6 Sol 74.8% Vals AI harness, v11 full set Sep 2026 Official leaderboard
14 Muse Wonder Full 73.3% Vals AI harness, v11 full set Oct 2026 Official leaderboard
15 Kimi K3 72.6% Vals AI harness, v11 full set Jul 2026 Official leaderboard
16 Grok 4.6 72.2% Vals AI harness, v11 full set Aug 2026 Official leaderboard
17 Tencent Hy4 preview 71.3% Aug 2026 Lab model card
18 GPT-5.6 Sol 71.1% Vals AI harness, v11 full set Jul 2026 Official leaderboard
19 Claude Haiku 5.5 70.4% Vals AI harness, v11 full set Oct 2026 Official leaderboard
20 Inkling 2 Small 70% Vals AI harness, v11 full set Nov 2026 Official leaderboard
21 Grok 4.7 69.3% Vals AI harness, v11 full set Sep 2026 Official leaderboard
22 MiMo-V2.6 Flash 69.3% Vals AI harness, v11 full set Sep 2026 Official leaderboard
23 DeepSeek V4.1 Flash 67.8% Vals AI harness, v11 full set Sep 2026 Official leaderboard
24 Step 5 Preview 67.4% Vals AI harness, v11 full set Sep 2026 Official leaderboard
25 Ember-1 67% Vals AI harness, v11 full set Sep 2026 Official leaderboard
26 Mistral Large 4 Preview 67% Vals AI harness, v11 full set Oct 2026 Official leaderboard
27 Muse Spark 1.2 64.8% Vals AI harness, v11 full set Aug 2026 Official leaderboard
28 DeepSeek V4 Flash 0731 64.4% Vals AI harness, v11 full set Jul 2026 Official leaderboard
29 Claude Opus 4.6 61.5% Feb 2026 Lab model card
30 GPT-5.6 Luna 61.5% Vals AI harness, v11 full set Jul 2026 Official leaderboard
31 GPT-6 Luna 61.5% Vals AI harness, v11 full set Sep 2026 Official leaderboard
32 Gemini 3.1 Pro 60% Vals AI harness, v11 full set Feb 2026 Official leaderboard
33 Gemini 3.6 Flash 58.5% Vals AI harness, v11 full set Jul 2026 Official leaderboard
34 Claude Sonnet 4.6 50.4% Feb 2026 Lab model card
35 Beam 1.0 42.6% Vals AI harness, v11 full set Oct 2026 Official leaderboard

Labs report scores for 10 models in their cards. Switch the scores above to rank them together.

6 models have held number one on Vals AI’s leaderboard since its release in April 2026 4 models have held number one with claims included since its release in April 2026

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

Google DeepMind OpenAI Moonshot AI Anthropic
Vals AI leaderboard 6 leaders
Claude Fable 5.2 83.7%
Lab claims Passed 85% in Jul 2026, by Claude Opus 5
Claude Fable 5.1 90.3%

Sources: Vals AI’s leaderboard and labs’ model cards

Anthropic’s own number for Claude Opus 5 passed 85% in July 2026

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

On Vals AI’s leaderboard Reported by the lab in its model card Best so far
0% 20% 40% 60% 80% 100% 2026 2027 Near the ceiling (85%) Claude Fable 5.1 Claude Fable 5 Claude Opus 4.6 0% 20% 40% 60% 80% 100% 2026 2027 Near the ceiling (85%) Claude Fable 5.1 Claude Fable 5

Sources: Vals AI’s leaderboard and labs’ model cards

5 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.

Anthropic
Google DeepMind
Tencent
Moonshot AI
Thinking Machines
DateLaunch postScoreAs the post writes it
Nov 5, 2026 Inkling 2 Small 70 BioMysteryBench
Oct 24, 2026 Kimi K3.1 Pro 80.6 BioMysteryBench (v11)
Sep 2, 2026 Gemini 3.8 Flash 88.8 Best of BioMysteryBench, Human Solvable / BioMysteryBench, Human Difficult
Aug 28, 2026 Tencent Hy4 preview 71.3 BioMysteryBench
Aug 13, 2026 Gemini 3.7 Flash 87.1 Best of BioMysteryBench, Human solvable / BioMysteryBench, Human difficult
Jul 24, 2026 Claude Opus 5 90.1 Best of Biology BioMysteryBench (hard) / Biology BioMysteryBench (human solved)
Jun 9, 2026 Claude Fable 5 83.9 Best of Biology BioMysteryBench (hard) / Biology BioMysteryBench (human solved)

11 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
Sep 28, 2026 Claude Sonnet 5.5 system card BioMysteryBench, Human Solvable 89.2% and Human Difficult 44.7%, no overall (2.2.2 and 8.17.1, Figure 8.17.A) Bio and chem risk
Sep 22, 2026 Claude Opus 5.5 system card BioMysteryBench, Human Solvable 89.3% and Human Difficult 50.0%, no overall, final model (2.2.3.1 and 8.17.1) Bio and chem risk
Sep 2, 2026 Gemini 3.8 Flash model card 88.8 BioMysteryBench, Human Solvable (results table, p.5). Human Difficult = 56.5% Capabilities
Sep 1, 2026 Claude Fable 5.1 system card 90.3 BioMysteryBench, 90.3% solvable and 44.1% difficult Capabilities
Aug 13, 2026 Gemini 3.7 Flash model card 87.1 BioMysteryBench, Human solvable (results table, p.5). Human difficult = 43.5% Capabilities
Jul 24, 2026 Claude Opus 5 system card BioMysteryBench, Human Solvable 90.1% and Human Difficult 49.4%, no overall (8.17.1) Capabilities
Jun 30, 2026 Claude Sonnet 5 system card BioMysteryBench, Human Solvable 0.82 and Human Difficult 0.35, no overall (8.14.1, Figure 8.14.6.A) Capabilities
May 28, 2026 Claude Opus 4.8 system card BioMysteryBench Verified, Human Solvable 80.4% and Human Difficult 40.0%, no overall (8.16.2) Capabilities
Apr 16, 2026 Claude Opus 4.7 system card BioMysteryBench, Verified (human-solvable) subset 78.9 and Hard subset 20.9, no overall, no extended thinking (8.13.1.2) Capabilities
Feb 17, 2026 Claude Sonnet 4.6 system card 50.4 Computational Biology, BioMysteryBench, no extended thinking (2.21.1) Capabilities
Feb 5, 2026 Claude Opus 4.6 system card 61.5 Computational Biology, BioMysteryBench, bash tool, no extended thinking, above the human expert baseline (2.22, 8.2.4.10) Capabilities

The lab's number comes out ahead in 4 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. 1 of them match, which usually means one side copied the other. Every benchmark.

Model Lab says Board says Gap
Gemini 3.8 Flash 88.8% Best of BioMysteryBench, Human Solvable / BioMysteryBench, Human Difficult 62.2% Leaderboard, Vals AI harness, v11 full set +27 pts
Claude Opus 5 90.1% Best of Biology BioMysteryBench (hard) / Biology BioMysteryBench (human solved) 79.3% Leaderboard, Vals AI harness, v11 full set +11 pts
Kimi K3.1 Pro 80.6% BioMysteryBench (v11) 75.6% Leaderboard, Vals AI harness, v11 full set +5.0 pts
Tencent Hy4 preview 71.3% BioMysteryBench 69.3% Leaderboard, Vals AI harness, v11 full set +2.0 pts
Inkling 2 Small 70% BioMysteryBench 70% Leaderboard, Vals AI harness, v11 full set Same

More in Science and reasoning

All 92 in Science and reasoning
  • GPQA Diamond In use 198 graduate-level 'Google-proof' multiple-choice questions in biology, chemistry and physics. Gemini 4 Carbon 96.5%
  • Terminal-Bench-Science In use 70 scientific research-workflow tasks across five science areas, run in Harbor containers. Hidden tests give the resolution rate. GPT-6 Astra 68.1%
  • FrontierScience In use Models solve physics, chemistry, and biology problems in an Olympiad track graded by short answer and a Research track graded by 10-point rubrics. Intern-S2-397B 87%
  • HiPhO In use Models solve theory problems from 2024 and 2025 physics olympiad exams, some with diagrams. Official marking schemes grade answers and steps, mapped to medal levels. Gemini 3 Pro 78.3%
  • LABBench2 In use 11 biology research sub-tasks (literature, figures, tables, patents, trials and more) with tool access. Macro-averaged accuracy. GPT-5.2 60.9%
  • SGI-Bench In use Models work through more than 1,000 expert samples in four scientist tasks: deep research, idea generation, dry and wet experiments, and experimental reasoning. Gemini 3 Pro 33.8%