CategoryKnowledge
MeasureChance-normalized average precision
Tasks83 protein-design library-ranking problems
DifficultyComputational protein design
Ranks held-out protein designs for experimental testing using wet-lab campaign measurements and a selection objective.
Protein Design library ranking ranking
2 models with a published Protein Design library ranking value, ordered by that value, highest first. Models the source has not scored on this benchmark are not listed — they are unmeasured, not last.
Models ranked by their published Anthropic protein-design library-ranking task value| Rank | Model | Provider | Chance-normalized average precision |
|---|
| 1 |  Claude Opus 5.5 | Anthropic | 56.0 |
| 2 | Claude Sonnet 5.5 | Anthropic | 54.8 |
Evidence key: Observed
Rows are ordered by the value Claude Opus 5.5 System Card published, highest first. The source does not state whether a higher value is the better result, so this page does not either: for a benchmark that measures a rate of failure, read the table from the bottom.
About Protein Design library ranking
What the benchmark measures and how its value is published, from the source’s own record.
What is Protein Design library ranking?
Protein Design library ranking (Anthropic protein-design library-ranking task) is a knowledge benchmark published by Claude Opus 5.5 System Card. Ranks held-out protein designs for experimental testing using wet-lab campaign measurements and a selection objective. Its evaluation set is described by the source as 83 protein-design library-ranking problems. The source rates its difficulty as computational protein design.
How is Protein Design library ranking scored?
Each model's result is published by Claude Opus 5.5 System Card as chance-normalized average precision. ModelScale prints that value to 1 decimal place, exactly as published, and never converts or rescales it. This page ranks the 2 models that carry a published value, highest value first; a model the source never scored is left out rather than ranked at zero. The source does not state whether a higher value is the better result, so neither does this page.
ModelScale