Creative biology
Models answer questions about unusual biology needed to engineer harmless biological agents, a weak proxy for novel weapon design.
SecureBio question set run by Anthropic. The questions are not public.
- 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
- Short answer A number, word or short string matched to a key.
Claude Opus 4.6 leads with 60.3% among models released by Feb 2026
Best score per model as labs report them in their own cards. Labs pick their own settings, so these compare less cleanly than independent runs.
The newest model with a score here came out in Feb 2026, and nobody has reported a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6 Astra
- Claude Opus 5.5
- GPT-6.1 Terra
- GPT-6.1 Sol
- and 16 more
- 1 Claude Opus 4.6 60.3%
- 2 Claude Sonnet 4.6 59.3%
- 3 Claude Opus 4.5 52.4%
- 4 Claude Sonnet 4.5 48.8%
- 5 Claude Opus 4.1 48%
- 6 Claude Opus 4 45%
- 7 Claude Sonnet 4 37%
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Claude Opus 4.6 | 60.3% | Feb 2026 | Model card | |
| 2 | Claude Sonnet 4.6 | 59.3% | Feb 2026 | Model card | |
| 3 | Claude Opus 4.5 | 52.4% | Nov 2025 | Model card | |
| 4 | Claude Sonnet 4.5 | 48.8% | Sep 2025 | Model card | |
| 5 | Claude Opus 4.1 | 48% | Aug 2025 | Model card | |
| 6 | Claude Opus 4 | 45% | May 2025 | Model card | |
| 7 | Claude Sonnet 4 | 37% | May 2025 | Model card |
5 models have held number one since its release in May 2025
Lab claims count from the day the card came out, independent runs from the model’s release.
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Labs report it only in system cards
We read the headline results of 479 launch posts from 51 labs, and none of them list this benchmark. Some system cards do, below.
7 system cards report it, from 1 lab
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.
| Date | Document | Score | As the document writes it | Section |
|---|---|---|---|---|
| 17 Feb 2026 | Claude Sonnet 4.6 system card | 59.3 | Creative biology automated evaluation, score 0.593 (6.2.2.4) | Bio and chem risk |
| 5 Feb 2026 | Claude Opus 4.6 system card | 60.3 | Creative Biology Automated Evaluations, score 0.603 (8.2.4.8) | Bio and chem risk |
| 24 Nov 2025 | Claude Opus 4.5 system card | 52.4 | Creative biology score 0.524 (7.2.4.5) | Bio and chem risk |
| 29 Sep 2025 | Claude Sonnet 4.5 system card | 48.8 | Creative biology mean score 0.488 (Figure 9.2.4.5.A). Text gives 0.48 ± 0.08 and best of 30 attempts 0.79 (9.2.4.5) | Bio and chem risk |
| 5 Aug 2025 | Claude Opus 4.1 system card addendum | 48 | Creative biology, score 0.48 ± 0.09 (6.3.1.1) | Bio and chem risk |
| 22 May 2025 | Claude Sonnet 4 system card | 37 | Creative biology, mean score 0.37 ± 0.08. Best of 30 attempts 0.74 (7.2.4.8) | Bio and chem risk |
| 22 May 2025 | Claude Opus 4 system card | 45 | Creative biology, mean score 0.45 ± 0.08. Best of 30 attempts 0.81 (7.2.4.8) | Bio and chem risk |
More 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%