# Anthropic protein-design library-ranking task (Protein Design library ranking)

> Ranks held-out protein designs for experimental testing using wet-lab campaign measurements and a selection objective.

Canonical page: https://benchlm.ai/benchmarks/proteindesignlibraryranking

- Category: [Knowledge](/knowledge)
- Last updated: September 27, 2026

## About Protein Design library ranking

- Year: 2026
- Tasks: 83 protein-design library-ranking problems
- Format: Chance-normalized average precision
- Difficulty: Computational protein design
- Paper: [Claude Opus 5.5 System Card](https://www-cdn.anthropic.com/fc1b44717c85dc068bc6ba5024219938094694bd/Claude%20Opus%205.5%20System%20Card.pdf)

Section 8.17.5 reports chance-normalized average precision across the current 83-problem version. Earlier system-card results use a different problem set and grader, so this lane remains distinct and display-only.

Protein Design library ranking is currently displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.

## Leaderboard (1 models)

| Rank | Model | Creator | Score |
|------|-------|---------|-------|
| 1 | [Claude Opus 5.5](/models/claude-opus-5-5) | Anthropic | 56.0% |

## FAQ

### What does Protein Design library ranking measure?

Ranks held-out protein designs for experimental testing using wet-lab campaign measurements and a selection objective.

### Which model scores highest on Protein Design library ranking?

Claude Opus 5.5 by Anthropic currently leads with a score of 56.0% on Protein Design library ranking.

### How many models are evaluated on Protein Design library ranking?

1 AI models have been evaluated on Protein Design library ranking on BenchLM.

### Does Protein Design library ranking affect BenchLM's overall score?

Not directly. Protein Design library ranking is still displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.