URSA - Expert-2026 Solv-2: leaderboard
URSA (Insilico Medicine, 2026) evaluates synthesis planning: each proposed route must end in purchasable building blocks, be structurally valid and pass the ChemCensor precedent-based plausibility check for every reaction (Solv-2). This board is the percent of the 100 novel drug-like URSA-expert-2026 targets solved at Solv-2 in the open URSA-minor-1.1.0-U2 configuration; higher is better.
Metric: Solv-2 solved targets (%; share of the 100 URSA-expert-2026 novel, not-yet-synthesized drug-like targets with stock-terminated routes proposed for those targets whose every reaction also passes the ChemCensor precedent-based plausibility check (Solv-2), best of up to 10 routes per target, URSA-minor-1.1.0-U2 configuration with ChemCensor v1.2.0 and the U2 USPTO precedent database, shared stock of 255,365 building blocks; default reasoning effort; higher is better). Source: arxiv.org. Saturation forecast: Around June 2027. 17 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 | 21 | #49 |
| 2 | Gemini 3.1 Pro (Preview) | 13 | #94 |
| 3 | Claude Opus 4.8 | 10 | #56 |
| 4 | Claude Opus 4.7 | 8 | #79 |
| 5 | Grok 4.1 | 7 | #476 |
| 6 | Claude Opus 4.6 | 5 | #102 |
| 7 | Kimi K2.5 | 5 | #256 |
| 8 | Claude Sonnet 4.6 | 3 | #161 |
| 9 | Claude Opus 4.5 | 3 | #156 |
| 10 | Qwen 3.5 397B A17B | 2 | #246 |
| 11 | GLM-5 | 2 | #260 |
| 12 | Claude Sonnet 4.5 | 1 | #250 |
| 13 | GPT-5.1 | 1 | #229 |
| 14 | GPT-5.4 | 0 | #123 |
| 15 | GPT-5.2 | 0 | #176 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=ursa-expert-2026-solv-2 · How It Works · Data refreshed daily, snapshot 2026-11-13.