Benchmark Almanac

SpatialBench-Long

Claude Opus 5.5 40.5% leads since September 2026
Not in model cards

Agents run long, multi-step analyses of real spatial biology datasets that need sustained biological reasoning.

Part of LatchBio's BioBench suite on benchmarks.bio. The V1 set has 21 evaluations. Release month is the GitHub repo creation month.

Made by
LatchBio
Released
May 2026
Kind
Genomics and life sciences
Level
Expert
Scored
Checked automatically
Answers
Agent task
Tasks
Public, plus a private test set
Strongest model tested
GPT-6 Astra #3 on the capability index
Last model added
October 2026
Scores refreshed
November 12th, 2026
27
Models scored on its own leaderboard

2 models share the top score of 40.5%

Best score per model from the benchmark's own leaderboard, read on November 12th, 2026. Where a model ran with several settings or harnesses, we keep its best.

27 models. Scroll sideways for the rest.

  1. 40.5 Claude Opus 5.5
  2. 40.5 GPT-6.1 Terra
  3. 39.9 GPT-6.1 Sol
  4. 39.3 GPT-6 Astra
  5. 37.5 GPT-5.6 Sol
  6. 36.5 Claude Sonnet 5.5
  7. 33.3 Claude Opus 4.8
  8. 32.1 GPT-5.5
  9. 31.6 GPT-6 Sol
  10. 30.9 Claude Opus 5
  11. 29.8 Claude Opus 4.7
  12. 29.8 Gemini 3.7 Flash
  13. 29.8 Claude Fable 5.1
  14. 28.6 Gemini 3.8 Flash
  15. 28.6 Gemini 4 Argon
  16. 27.4 Grok 4.6
  17. 26.2 DeepSeek V4.1 Flash
  18. 25.0 GPT-5.6 Luna
  19. 24.4 Claude Sonnet 5
  20. 24.4 Grok 4.7
  21. 23.2 GPT-6 Luna
  22. 22.6 Claude Sonnet 4.6
  23. 22.6 Kimi K3
  24. 20.8 Mistral Large 4
  25. 18.4 GLM-5.3
  26. 15.9 Nemotron 3 Ultra
  27. 4.76 Nemotron 3.5 Lightning 30B-A3B
  1. 1 Claude Opus 5.5 LSVP 40.5%
  2. 2 GPT-6.1 Terra 40.5%
  3. 3 GPT-6.1 Sol 39.9%
  4. 4 GPT-6 Astra 39.3%
  5. 5 GPT-5.6 Sol 37.5%
  6. 6 Claude Sonnet 5.5 36.5%
  7. 7 Claude Opus 4.8 33.3%
  8. 8 GPT-5.5 32.1%
  9. 9 GPT-6 Sol 31.6%
  10. 10 Claude Opus 5 30.9%
  11. 11 Claude Opus 4.7 29.8%
  12. 12 Gemini 3.7 Flash 29.8%
  13. 13 Claude Fable 5.1 LSVP 29.8%
  14. 14 Gemini 3.8 Flash 28.6%
  15. 15 Gemini 4 Argon 28.6%
  16. 16 Grok 4.6 27.4%
  17. 17 DeepSeek V4.1 Flash 26.2%
  18. 18 GPT-5.6 Luna 25%
  19. 19 Claude Sonnet 5 24.4%
  20. 20 Grok 4.7 24.4%
  21. 21 GPT-6 Luna 23.2%
  22. 22 Claude Sonnet 4.6 22.6%
  23. 23 Kimi K3 22.6%
  24. 24 Mistral Large 4 20.8%
  25. 25 GLM-5.3 18.4%
  26. 26 Nemotron 3 Ultra 15.9%
  27. 27 Nemotron 3.5 Lightning 30B-A3B 4.76%

Source: Official SpatialBench-Long leaderboard

#ModelScoreSettingDateSource
1 Claude Opus 5.5 40.5% LSVP, Claude Code Sep 2026 Official leaderboard
2 GPT-6.1 Terra 40.5% Pi Oct 2026 Official leaderboard
3 GPT-6.1 Sol 39.9% Codex Sep 2026 Official leaderboard
4 GPT-6 Astra 39.3% Codex Sep 2026 Official leaderboard
5 GPT-5.6 Sol 37.5% Pi Jul 2026 Official leaderboard
6 Claude Sonnet 5.5 36.5% Claude Code Sep 2026 Official leaderboard
7 Claude Opus 4.8 33.3% Pi May 2026 Official leaderboard
8 GPT-5.5 32.1% Pi Apr 2026 Official leaderboard
9 GPT-6 Sol 31.6% Pi Sep 2026 Official leaderboard
10 Claude Opus 5 30.9% Claude Code Jul 2026 Official leaderboard
11 Claude Opus 4.7 29.8% Claude Code Apr 2026 Official leaderboard
12 Gemini 3.7 Flash 29.8% Pi Aug 2026 Official leaderboard
13 Claude Fable 5.1 29.8% LSVP, Pi Sep 2026 Official leaderboard
14 Gemini 3.8 Flash 28.6% Pi Sep 2026 Official leaderboard
15 Gemini 4 Argon 28.6% Pi Sep 2026 Official leaderboard
16 Grok 4.6 27.4% Grok Build Aug 2026 Official leaderboard
17 DeepSeek V4.1 Flash 26.2% Pi Sep 2026 Official leaderboard
18 GPT-5.6 Luna 25% Codex Jul 2026 Official leaderboard
19 Claude Sonnet 5 24.4% Pi Jun 2026 Official leaderboard
20 Grok 4.7 24.4% Pi Sep 2026 Official leaderboard
21 GPT-6 Luna 23.2% Pi Sep 2026 Official leaderboard
22 Claude Sonnet 4.6 22.6% Pi Feb 2026 Official leaderboard
23 Kimi K3 22.6% Pi Jul 2026 Official leaderboard
24 Mistral Large 4 20.8% Pi Oct 2026 Official leaderboard
25 GLM-5.3 18.4% Pi Aug 2026 Official leaderboard
26 Nemotron 3 Ultra 15.9% Pi Jun 2026 Official leaderboard
27 Nemotron 3.5 Lightning 30B-A3B 4.76% Pi Aug 2026 Official leaderboard

5 models have held number one on the official leaderboard since its release in May 2026

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

OpenAI Anthropic
Own leaderboard 5 leaders
Claude Opus 5.5 40.5%

Source: Official SpatialBench-Long leaderboard

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 GPT-5.6 Sol Claude Opus 4.8 Claude Sonnet 4.6 0% 20% 40% 60% 80% 100% 2026 2027 Near the ceiling (85%) Claude Opus 5.5 Claude Opus 4.8

Source: Official SpatialBench-Long leaderboard

No lab model card reports it yet

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

More in Biology and chemistry

All 108 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%