MolViBench - Level 5 Create: leaderboard

The Level 5 slice of MolViBench (2026), a molecular vibe coding benchmark: RDKit tasks on creating: end-to-end cheminformatics pipeline construction and system-level integration (64 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 64 Level 5 tasks (Create), 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
1GPT-5.2 Codex9.4#182
2Gemini 3 Pro7.8#145
3DeepSeek V3.2Non-reasoning7.8#419
4Claude Opus 4.66.2#113
5GPT-5.3 Codex6.2#124
6Kimi K2.54.7#266
7MiniMax-M2.53.1#600

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-5-create · How It Works · Data refreshed daily, snapshot 2026-11-21.