Molmo2-ER Release - VSI-Bench: leaderboard
Standalone Molmo2-ER embodied reasoning backbone;13spatial/pointing/QA tasks plus mean. Source-specific points and judge protocols retained. Action-policy fine-tunes and embodiments in other tables are distinct models, not this backbone. Metric: Benchmark score (%). Higher is better. Source: https://arxiv.org/html/2605.02881v1
Metric: Benchmark score (%). Source: arxiv.org. Saturation forecast: Around January 2027. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 | 52.9 | #95 |
| 2 | Gemini 2.5 Pro | 51.1 | #151 |
| 3 | GPT-5 Mini | 46.2 | #182 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #7 Gemini 4 Argon.
Interactive version: theaggregate.ai/benchmark?slug=molmo2-er-release-vsi-bench · How It Works · Data refreshed daily, snapshot 2026-10-17.