Protocols (Anthropic life sciences)
Models find and fix errors in molecular biology protocols and answer practical questions about applying real laboratory protocols.
Reported as two variants, Troubleshooting and Understanding (V2), with no overall score. Benchling built the Understanding set. From the Claude Opus 5.5 system card it is one of three evaluations Anthropic uses for CB-1 capability, with VCT and BioMysteryBench, replacing the Long-Form Virology Tasks.
- Made by
- Anthropic (with Benchling)
- Released
- July 2026
- Kind
- Lab protocols
- Level
- Expert
- Scored
- Checked automatically
- Answers
- Short answer
- Tasks
- Kept private
- Strongest model tested
- Claude Opus 4.8 #21 on the capability index
- Last in a system card
- October 2026
- Last model added
- October 2026
Claude Sonnet 5 leads with 62% among models released by June 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 June 2026, and nobody has reported a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- Claude Fable 5.2
-
Gemini 4 Carbon - GPT-6 Astra
- Claude Opus 5.5
- and 16 more
- 1 Claude Sonnet 5 62%
- 2 Claude Opus 4.8 59.6%
- 3 Claude Opus 4.7 51.8%
Source: Labs’ model cards
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Claude Sonnet 5 | 62% | Jun 2026 | Lab model card | |
| 2 | Claude Opus 4.8 | 59.6% | May 2026 | Lab model card | |
| 3 | Claude Opus 4.7 | 51.8% | Apr 2026 | Lab model card |
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Labs’ model cards
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Labs’ model cards
Labs report it only in system cards
We read the headline results of 522 launch posts from 52 labs, and none of them list this benchmark. Some system cards do, below.
8 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 |
|---|---|---|---|---|
| Oct 7, 2026 | Claude Haiku 5.5 system card | Protocols, Troubleshooting 58.0 and Understanding (V2) 64.1, no overall (8.13.6, CB-1 evaluation in 2.2.2) | Bio and chem risk | |
| Sep 28, 2026 | Claude Sonnet 5.5 system card | Protocols, Troubleshooting 67.3% and Understanding (V2) 66.6%, no overall (2.2.2 and 8.17.8, Figure 8.17.A) | Bio and chem risk | |
| Sep 22, 2026 | Claude Opus 5.5 system card | Protocols, Troubleshooting 73.7% and Understanding (V2) 69.0%, no overall, final model (2.2.3.1 and 8.17.8) | Bio and chem risk | |
| Sep 1, 2026 | Claude Fable 5.1 system card | Protocols, Troubleshooting 70.2% and Understanding 77.2%, no overall, Mythos 5.1 (8.19.6) | Other | |
| Jul 24, 2026 | Claude Opus 5 system card | Protocols, Troubleshooting 61.1% and Understanding 78.4%, no overall (8.17.6) | Capabilities | |
| Jun 30, 2026 | Claude Sonnet 5 system card | 62 | Protocol troubleshooting, with web search, score 0.62. Only the Troubleshooting variant is reported (8.14.6, Figure 8.14.6.A) | Capabilities |
| May 28, 2026 | Claude Opus 4.8 system card | 59.6 | Protocol troubleshooting with bash and web search tools. Only the Troubleshooting part is reported (8.16.7) | Capabilities |
| Apr 16, 2026 | Claude Opus 4.7 system card | 51.8 | Protocol troubleshooting with bash and web search tools. Only the troubleshooting variant is reported (8.13.6) | Capabilities |
More 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%