Benchmark Almanac

OncologyBench

Claude Opus 5.5 50.7% leads since Oct 2026 Not in model cards

Separate agents run analyses of cancer research data, and a final agent makes a decision on detection, diagnosis or treatment. Strict scoring needs full marks on both parts.

Part of LatchBio's BioBench suite on benchmarks.bio. The V1 set has 25 evaluations. Release month is the GitHub repo creation month.

Made by LatchBio Released Oct 2026 Genomics and life sciences All tasks public Last model added Oct 2026 Scores refreshed 2 Nov 2026
Level
Expert Specialists with years of training, often over hours of work
Scored
Graded by a model A language model scores each response, usually against a rubric.
Answers
Agent task Many steps in a terminal, browser, desktop, app, game or robot.
latchbio/oncologybench 4 stars Leaderboard benchmarks.bio
30
Models scored on its own leaderboard

Claude Opus 5.5 leads with 50.7%

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. Every run is in the next chart.

30 models. Scroll sideways for the rest.

  1. 50.7 Claude Opus 5.5
  2. 44.0 Claude Fable 5.1
  3. 42.7 Claude Sonnet 5.5
  4. 40.0 Claude Opus 5
  5. 33.3 Gemini 4 Argon
  6. 24.0 Kimi K3
  7. 24.0 GPT-6.1 Sol
  8. 24.0 Muse Wonder
  9. 22.7 Mistral Large 4
  10. 21.3 GPT-6 Astra
  11. 20.0 GPT-6 Sol
  12. 18.7 GPT-5.6 Sol
  13. 18.7 Grok 4.6
  14. 18.7 Grok 4.7
  15. 17.3 GLM-5.3
  16. 16.0 Claude Opus 4.8
  17. 16.0 GPT-6 Luna
  18. 16.0 GPT-6.1 Terra
  19. 14.7 GPT-5.5
  20. 14.7 Claude Sonnet 5
  21. 14.7 Gemini 3.7 Flash
  22. 14.7 DeepSeek V4.1 Flash
  23. 13.3 Gemini 3.8 Flash
  24. 12.0 Muse Spark Open
  25. 10.7 GPT-5.6 Luna
  26. 8.00 Mercury 3
  27. 5.33 Claude Sonnet 4.6
  28. 1.33 Nemotron 3 Ultra
  29. 1.33 Beam 1.0
  30. 0.0 Nemotron 3.5 Lightning 30B-A3B
  1. 1 Claude Opus 5.5 LSVP 50.7%
  2. 2 Claude Fable 5.1 LSVP 44%
  3. 3 Claude Sonnet 5.5 LSVP 42.7%
  4. 4 Claude Opus 5 40%
  5. 5 Gemini 4 Argon 33.3%
  6. 6 Kimi K3 24%
  7. 7 GPT-6.1 Sol 24%
  8. 8 Muse Wonder 24%
  9. 9 Mistral Large 4 22.7%
  10. 10 GPT-6 Astra 21.3%
  11. 11 GPT-6 Sol 20%
  12. 12 GPT-5.6 Sol 18.7%
  13. 13 Grok 4.6 18.7%
  14. 14 Grok 4.7 18.7%
  15. 15 GLM-5.3 17.3%
  16. 16 Claude Opus 4.8 16%
  17. 17 GPT-6 Luna 16%
  18. 18 GPT-6.1 Terra 16%
  19. 19 GPT-5.5 14.7%
  20. 20 Claude Sonnet 5 14.7%
  21. 21 Gemini 3.7 Flash 14.7%
  22. 22 DeepSeek V4.1 Flash 14.7%
  23. 23 Gemini 3.8 Flash 13.3%
  24. 24 Muse Spark Open 12%
  25. 25 GPT-5.6 Luna 10.7%
  26. 26 Mercury 3 8.00%
  27. 27 Claude Sonnet 4.6 5.33%
  28. 28 Nemotron 3 Ultra 1.33%
  29. 29 Beam 1.0 1.33%
  30. 30 Nemotron 3.5 Lightning 30B-A3B 0%
#ModelScoreSettingDateSource
1 Claude Opus 5.5 50.7% LSVP, Pi Sep 2026 Official leaderboard
2 Claude Fable 5.1 44% LSVP, Pi Sep 2026 Official leaderboard
3 Claude Sonnet 5.5 42.7% LSVP, Claude Code Sep 2026 Official leaderboard
4 Claude Opus 5 40% Pi Jul 2026 Official leaderboard
5 Gemini 4 Argon 33.3% Pi Sep 2026 Official leaderboard
6 Kimi K3 24% Pi Jul 2026 Official leaderboard
7 GPT-6.1 Sol 24% Pi Sep 2026 Official leaderboard
8 Muse Wonder 24% Pi Oct 2026 Official leaderboard
9 Mistral Large 4 22.7% Pi Oct 2026 Official leaderboard
10 GPT-6 Astra 21.3% Pi Sep 2026 Official leaderboard
11 GPT-6 Sol 20% Codex Sep 2026 Official leaderboard
12 GPT-5.6 Sol 18.7% Pi Jul 2026 Official leaderboard
13 Grok 4.6 18.7% Pi Aug 2026 Official leaderboard
14 Grok 4.7 18.7% Pi Sep 2026 Official leaderboard
15 GLM-5.3 17.3% Pi Aug 2026 Official leaderboard
16 Claude Opus 4.8 16% Claude Code May 2026 Official leaderboard
17 GPT-6 Luna 16% Codex Sep 2026 Official leaderboard
18 GPT-6.1 Terra 16% Pi Oct 2026 Official leaderboard
19 GPT-5.5 14.7% Pi Apr 2026 Official leaderboard
20 Claude Sonnet 5 14.7% Pi Jun 2026 Official leaderboard
21 Gemini 3.7 Flash 14.7% Pi Aug 2026 Official leaderboard
22 DeepSeek V4.1 Flash 14.7% Pi Sep 2026 Official leaderboard
23 Gemini 3.8 Flash 13.3% Pi Sep 2026 Official leaderboard
24 Muse Spark Open 12% Pi Oct 2026 Official leaderboard
25 GPT-5.6 Luna 10.7% Codex Jul 2026 Official leaderboard
26 Mercury 3 8.00% Pi Oct 2026 Official leaderboard
27 Claude Sonnet 4.6 5.33% Claude Code Feb 2026 Official leaderboard
28 Nemotron 3 Ultra 1.33% Pi Jun 2026 Official leaderboard
29 Beam 1.0 1.33% Pi Oct 2026 Official leaderboard
30 Nemotron 3.5 Lightning 30B-A3B 0% Pi Aug 2026 Official leaderboard

Scores at each effort level

Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.

Lowest effort to highest, light to dark
Claude Opus 5.5 Pi
LSVP
50.7%
Claude Sonnet 5.5 Claude Code
LSVP
42.7%
Claude Opus 5.5 Claude Code
LSVP
40%
Claude Sonnet 5.5 Pi
LSVP
36%
ModelEffortHarnessScoreCost per taskOutput tokens
Claude Opus 5.5 LSVP Pi 50.7%
Claude Fable 5.1 LSVP Pi 44%
Claude Sonnet 5.5 LSVP Claude Code 42.7%
Claude Opus 5 Pi 40%
Claude Opus 5.5 LSVP Claude Code 40%
Claude Opus 5.5 Claude Code 37.3%
Claude Opus 5 Claude Code 36%
Claude Sonnet 5.5 LSVP Pi 36%
Gemini 4 Argon Pi 33.3%
Claude Sonnet 5.5 Claude Code 32%
Claude Opus 5.5 Pi 29.3%
Claude Sonnet 5.5 Pi 29.3%
Kimi K3 Pi 24%
GPT-6.1 Sol Pi 24%
Muse Wonder Pi 24%
Mistral Large 4 Pi 22.7%
GPT-6 Astra Pi 21.3%
GPT-6.1 Sol Codex 21.3%
GPT-6 Sol Codex 20%
GPT-5.6 Sol Pi 18.7%
Grok 4.6 Pi 18.7%
GPT-6 Astra Codex 18.7%
Grok 4.7 Pi 18.7%
GPT-6 Sol Pi 18.7%
Grok 4.6 Grok Build 17.3%
GLM-5.3 Pi 17.3%
Claude Opus 4.8 Claude Code 16%
Claude Opus 4.8 Pi 16%
GPT-6 Luna Codex 16%
GPT-6.1 Terra Pi 16%
GPT-5.5 Pi 14.7%
Claude Sonnet 5 Pi 14.7%
GPT-5.6 Sol Codex 14.7%
Gemini 3.7 Flash Pi 14.7%
DeepSeek V4.1 Flash Pi 14.7%
Gemini 3.8 Flash Pi 13.3%
Claude Sonnet 5 Claude Code 12%
GPT-6 Luna Pi 12%
Muse Spark Open Pi 12%
GPT-5.6 Luna Codex 10.7%
GPT-5.6 Luna Pi 10.7%
GPT-6.1 Terra Codex 9.33%
Mercury 3 Pi 8.00%
Claude Sonnet 4.6 Claude Code 5.33%
Claude Sonnet 4.6 Pi 2.67%
Nemotron 3 Ultra Pi 1.33%
Beam 1.0 Pi 1.33%
Nemotron 3.5 Lightning 30B-A3B Pi 0%

Number one over time

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

Anthropic
Own leaderboard 1 leader
Claude Opus 5.5 50.7%

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 Kimi K3 GPT-5.6 Sol GPT-5.5 Claude Sonnet 4.6 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%