DK-VLM VSI-Bench - Route Planning: leaderboard
DK-VLM v1; LoRA/geometric attention fine-tunes remain distinct from Qwen2.5VL7BInstruct and Qwen3VL8BInstruct backbones. Frozen VGGT-Omega 1B frontend; official shared six-benchmark GeoThinker evaluation settings. Comparison baseline scores are source-reported, starred GeoThinker values reproduced by authors; no automatic primary-board fusion. Metric: Accuracy (%). Higher is better. Source: https://arxiv.org/html/2610.22356v1
Metric: Accuracy (%). Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 11 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | DK-VLM-8B | 60.31 | |
| 2 | DK-VLM-7B | 56.19 | |
| 3 | Q-GeoMem-7B | 50 | |
| 4 | GeoThinker-8B | 49.48 | |
| 5 | VG-LLM-8B | 47.4 | |
| 6 | GeoThinker-7B | 46.39 | |
| 7 | VLM-3R-7B | 45.4 | |
| 8 | Cambrian-S-7B | 41.8 | |
| 9 | SpatialStack-5B | 41.2 | |
| 10 | Qwen3-VL-8B (DK-VLM paper baseline) | 35 |
Interactive version: theaggregate.ai/benchmark?slug=dk-vlm-vsi-bench-route-planning · How It Works · Data refreshed daily, snapshot 2026-10-17.