VLAN - Matched 200 Trajectories - BVI: leaderboard
Independent evaluation cohort. S4T7=30real-worldtrajectories; language/actionfamilymeans andharmonicAccSR. A4T14=fixed200testtrajectories, 2/scene, seed42, nineVLMs, identicalbudget; notinterchangeablewith1931fullsplit. FullsplitDroneSRact/GRinconsistency held separately. Metric: Native success score (0–1). Source: https://arxiv.org/html/2611.18346v1
Metric: Native success score (0–1). Source: arxiv.org. Saturation forecast: Around February 2027. 8 models tracked. Updated 17 Nov 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4o | 0.425 | #754 |
| 2 | Qwen 3 VL 8B Instruct | 0.425 | #1025 |
| 3 | InternVL3.5-8B | 0.355 | #1588 |
| 4 | Qwen 2.5 VL 7B Instruct | 0.255 | #1893 |
| 5 | Qwen 2.5 VL 3B Instruct | 0.24 | #2648 |
| 6 | Llama 3.2 11B Instruct | 0.175 | #5226 |
| 7 | Molmo2-8B | 0.13 | #1629 |
| 8 | Idefics3-8B-Llama3 | 0.115 | #5775 |
No result here: #1 Claude Fable 5.2, #2 GPT-6.2 Astra, #3 Gemini 4 Carbon.
Interactive version: theaggregate.ai/benchmark?slug=vlan-matched-200-trajectories-bvi · How It Works · Data refreshed daily, snapshot 2026-11-17.