MolViBench - Level 3 Analyze: leaderboard

The Level 3 slice of MolViBench (2026), a molecular vibe coding benchmark: RDKit tasks on analyzing: multi-step chemical reasoning that needs domain knowledge (72 tasks). The score is Pass@1, the percent of tasks whose generated Python code reproduces the reference output, with each model generating once without execution feedback; higher is better.

Metric: Pass@1 (%) on the 72 Level 3 tasks (Analyze), direct generation in one pass with no execution feedback, RDKit molecular coding tasks answered by generated Python code, official APIs, temperature 0, at most 4,096 output tokens; higher is better. Source: arxiv.org. Saturation forecast: Around 2029. 7 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1Claude Opus 4.6Thinking30.6#113
2MiniMax-M2.526.4#600
3GPT-5.2 Codex23.6#182
4Gemini 3 Pro19.4#145
5DeepSeek V3.2Thinking19.4#419
6GPT-5.3 Codex16.7#124
7Kimi K2.56.9#266

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=molvibench-level-3-analyze · How It Works · Data refreshed daily, snapshot 2026-11-21.