Anthropic de novo protein-binder design evaluation (De novo protein-binder design)
We show this table for reference; we do not rank on it.
Designs diverse miniprotein binders against target protein structures in a GPU-equipped research sandbox.
Benchmark score on De novo protein-binder design — November 13, 2026
We compile the De novo protein-binder design rows from provider self-reports. Claude Mythos 5.2 leads the table at 85.1%, followed by Claude Opus 5.5 (82.6%) and Claude Sonnet 5.5 (82.3%). We do not use these results to rank models overall.
Claude Mythos 5.2
Anthropic
Reported by AnthropicClaude Opus 5.5
Anthropic
Reported by AnthropicClaude Sonnet 5.5
Anthropic
Reported by Anthropic4 modelsKnowledgeCurrentDisplay onlyUpdated November 13, 2026
Benchmark score table (4 models)
All 4 results on this page were reported by the labs that built the models.
| Rank | Model / configuration | Score | Who ran it | Parameters (B) | Open / closed |
|---|---|---|---|---|---|
| 1 | Claude Mythos 5.2AnthropicReported by Anthropic | 85.1% | Reported by Anthropic | Not reported | Closed |
| 2 | Claude Opus 5.5AnthropicReported by Anthropic | 82.6% | Reported by Anthropic | Not reported | Closed |
| 3 | Claude Sonnet 5.5AnthropicReported by Anthropic | 82.3% | Reported by Anthropic | Not reported | Closed |
| 4 | Claude Mythos 5.1AnthropicReported by Anthropic | 79.1% | Reported by Anthropic | Not reported | Closed |
- Lab-reported
- Benchmark-run
- Independent
Among the reported De novo protein-binder design rows, Claude Mythos 5.2 is first at 85.1%. The third row is 2.8 points behind. The broader top-10 range is 6.0 points, so many of the published results sit in a relatively narrow band.
4 models have been evaluated on De novo protein-binder design. The benchmark falls in the Knowledge category. De novo protein-binder design is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.
About De novo protein-binder design
Year
2026
Tasks
15 protein targets
Format
Composite structural design score
Difficulty
Long-horizon computational protein design
Section 8.17.6 reports the mean over 15 protein targets and five independent runs per target. It is display-only, model-plus-tool-system evidence.
Freshness and provenance
Version
De novo protein-binder design 2026
Refresh cadence
Quarterly
Staleness state
Current
Question availability
Public benchmark set
BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.
Questions
What does De novo protein-binder design measure?
Designs diverse miniprotein binders against target protein structures in a GPU-equipped research sandbox.
Which model scores highest on De novo protein-binder design?
Claude Mythos 5.2 by Anthropic currently leads with a score of 85.1% on De novo protein-binder design.
How many models are evaluated on De novo protein-binder design?
4 AI models have published results on De novo protein-binder design in the BenchLM catalog.
Compare top models on De novo protein-binder design
Know when it’s worth switching models
The model to choose, the cheaper alternative, and the release we would wait on.
Read a sample issueJoin 7,000+ subscribers.
One email each week. Unsubscribe anytime.