Benchmark Almanac

EpiBench

Claude Opus 5.5 39.6% leads since Sep 2026 Not in model cards

Agents analyze real epigenomics datasets in a sandbox with standard tools. A deterministic grader checks the biological result.

Part of LatchBio's BioBench suite on benchmarks.bio. The V1 set has 106 evaluations. Release month is the GitHub repo creation month.

Made by LatchBio Released Jun 2026 Genomics and life sciences Public tasks plus a private test set Last model added Oct 2026 Scores refreshed 2 Nov 2026
Level
Expert Specialists with years of training, often over hours of work
Scored
Checked automatically A program checks the result against an answer key, unit tests or the expected final state.
Answers
Agent task Many steps in a terminal, browser, desktop, app, game or robot.
latchbio/epibench 8 stars Leaderboard benchmarks.bio
27
Models scored on its own leaderboard

Claude Opus 5.5 leads with 39.6%

Best score per model from the benchmark's own leaderboard, read on 2 Nov 2026. Where a model ran with several settings or harnesses, we keep its best.

27 models. Scroll sideways for the rest.

  1. 39.6 Claude Opus 5.5
  2. 31.8 Claude Fable 5.1
  3. 30.2 Claude Sonnet 5.5
  4. 28.9 Gemini 4 Argon
  5. 28.3 GPT-5.5
  6. 28.3 GPT-6 Astra
  7. 28.3 GPT-6.1 Sol
  8. 25.8 Claude Opus 4.8
  9. 25.8 GPT-5.6 Sol
  10. 25.5 DeepSeek V4.1 Flash
  11. 25.2 Claude Opus 5
  12. 23.6 Grok 4.6
  13. 23.3 GPT-6.1 Terra
  14. 22.6 Claude Opus 4.7
  15. 22.0 Gemini 3.7 Flash
  16. 22.0 GPT-6 Sol
  17. 21.7 Claude Sonnet 5
  18. 21.4 Gemini 3.8 Flash
  19. 20.8 Grok 4.7
  20. 20.8 Mistral Large 4
  21. 19.8 GPT-5.6 Luna
  22. 19.2 GPT-6 Luna
  23. 18.9 Claude Sonnet 4.6
  24. 15.7 Kimi K3
  25. 12.9 GLM-5.3
  26. 11.3 Nemotron 3 Ultra
  27. 6.92 Nemotron 3.5 Lightning 30B-A3B
  1. 1 Claude Opus 5.5 LSVP 39.6%
  2. 2 Claude Fable 5.1 LSVP 31.8%
  3. 3 Claude Sonnet 5.5 30.2%
  4. 4 Gemini 4 Argon 28.9%
  5. 5 GPT-5.5 28.3%
  6. 6 GPT-6 Astra 28.3%
  7. 7 GPT-6.1 Sol 28.3%
  8. 8 Claude Opus 4.8 25.8%
  9. 9 GPT-5.6 Sol 25.8%
  10. 10 DeepSeek V4.1 Flash 25.5%
  11. 11 Claude Opus 5 25.2%
  12. 12 Grok 4.6 23.6%
  13. 13 GPT-6.1 Terra 23.3%
  14. 14 Claude Opus 4.7 22.6%
  15. 15 Gemini 3.7 Flash 22%
  16. 16 GPT-6 Sol 22%
  17. 17 Claude Sonnet 5 21.7%
  18. 18 Gemini 3.8 Flash 21.4%
  19. 19 Grok 4.7 20.8%
  20. 20 Mistral Large 4 20.8%
  21. 21 GPT-5.6 Luna 19.8%
  22. 22 GPT-6 Luna 19.2%
  23. 23 Claude Sonnet 4.6 18.9%
  24. 24 Kimi K3 15.7%
  25. 25 GLM-5.3 12.9%
  26. 26 Nemotron 3 Ultra 11.3%
  27. 27 Nemotron 3.5 Lightning 30B-A3B 6.92%
#ModelScoreSettingDateSource
1 Claude Opus 5.5 39.6% LSVP, Claude Code Sep 2026 Official leaderboard
2 Claude Fable 5.1 31.8% LSVP, Pi Sep 2026 Official leaderboard
3 Claude Sonnet 5.5 30.2% Pi Sep 2026 Official leaderboard
4 Gemini 4 Argon 28.9% Pi Sep 2026 Official leaderboard
5 GPT-5.5 28.3% Pi Apr 2026 Official leaderboard
6 GPT-6 Astra 28.3% Pi Sep 2026 Official leaderboard
7 GPT-6.1 Sol 28.3% Pi Sep 2026 Official leaderboard
8 Claude Opus 4.8 25.8% Pi May 2026 Official leaderboard
9 GPT-5.6 Sol 25.8% Codex Jul 2026 Official leaderboard
10 DeepSeek V4.1 Flash 25.5% Pi Sep 2026 Official leaderboard
11 Claude Opus 5 25.2% Pi Jul 2026 Official leaderboard
12 Grok 4.6 23.6% Pi Aug 2026 Official leaderboard
13 GPT-6.1 Terra 23.3% Pi Oct 2026 Official leaderboard
14 Claude Opus 4.7 22.6% Pi Apr 2026 Official leaderboard
15 Gemini 3.7 Flash 22% Pi Aug 2026 Official leaderboard
16 GPT-6 Sol 22% Pi Sep 2026 Official leaderboard
17 Claude Sonnet 5 21.7% Claude Code Jun 2026 Official leaderboard
18 Gemini 3.8 Flash 21.4% Pi Sep 2026 Official leaderboard
19 Grok 4.7 20.8% Pi Sep 2026 Official leaderboard
20 Mistral Large 4 20.8% Pi Oct 2026 Official leaderboard
21 GPT-5.6 Luna 19.8% Codex Jul 2026 Official leaderboard
22 GPT-6 Luna 19.2% Pi Sep 2026 Official leaderboard
23 Claude Sonnet 4.6 18.9% Pi Feb 2026 Official leaderboard
24 Kimi K3 15.7% Pi Jul 2026 Official leaderboard
25 GLM-5.3 12.9% Pi Aug 2026 Official leaderboard
26 Nemotron 3 Ultra 11.3% Pi Jun 2026 Official leaderboard
27 Nemotron 3.5 Lightning 30B-A3B 6.92% Pi Aug 2026 Official leaderboard

3 models have held number one since its release in Jun 2026

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

OpenAI Anthropic
Own leaderboard 3 leaders
Claude Opus 5.5 39.6%

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 Best so far
0% 20% 40% 60% 80% 100% 2026 2027 Near the ceiling (85%) Claude Opus 5.5 Claude Fable 5.1 GPT-5.5 Claude Opus 4.7 0% 20% 40% 60% 80% 100% 2026 2027 Near the ceiling (85%) Claude Opus 5.5 Claude Fable 5.1

No lab model card reports it yet

We read the headline results of 479 launch posts from 51 labs, and none of them list this benchmark.

More in Biology and chemistry

All 96 in Biology and chemistry
  • 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%
  • 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-6 Astra 37.1%
  • 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-6 Astra 60.3%
  • MolecularIQ In use Models answer questions about molecular graphs, such as counting, indexing and constrained generation, and a symbolic checker verifies each answer. Intern-S2-397B 62.4%
  • TOMG-Bench In use Models generate molecules from natural language requests in three tasks: molecule editing, molecule optimization and customized generation. Each output is checked for validity and for meeting the requested properties. Intern-S2-397B 66.8%
  • BABE Fading Models reason from experimental results in peer-reviewed biology papers to draw conclusions, with causal and cross-scale inference. Seed2.0 Lite (0428) 57.9%