DK-VLM VSI-Bench - Overall: 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: Official aggregate score (%). Higher is better. Source: https://arxiv.org/html/2610.22356v1
Metric: Official aggregate score (%). Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 11 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | DK-VLM-8B | 74.33 | |
| 2 | GeoThinker-8B | 72.6 | |
| 3 | Q-GeoMem-7B | 71.2 | |
| 4 | DK-VLM-7B | 69.42 | |
| 5 | GeoThinker-7B | 68.5 | |
| 6 | Cambrian-S-7B | 67.5 | |
| 7 | SpatialStack-5B | 67.5 | |
| 8 | VG-LLM-8B | 62.2 | |
| 9 | VLM-3R-7B | 60.9 | |
| 10 | Qwen3-VL-8B (DK-VLM paper baseline) | 57.9 |
Interactive version: theaggregate.ai/benchmark?slug=dk-vlm-vsi-bench-overall · How It Works · Data refreshed daily, snapshot 2026-10-17.