List your company

Video Spatial Reasoning on STI-Bench

41.4AccuracyGemini-2.5-pro
28.50431.85235.238.548Oct 10, 2025

Evaluation results

MethodLinks
Gemini-2.5-pro2025.1041.4
Qwen2.5-VL-72BParameters=72B2025.1040.7
GPT-52025.1039.3
InternVL3.5-38BParameters=38B2025.1039.2
SpaceVista-7BParameters=7B, RL=true2025.1038.2
SpaceR-7BParameters=7B2025.1037
SpaceVista-7BParameters=7B2025.1035.9
Qwen2.5-VL-7BFine-tuned on=SpaceVista-1M2025.1035
InternVL3.5-8BParameters=8B2025.1033.2
Qwen2.5-VL-7BParameters=7B2025.1032.1
VILASR-7BParameters=7B2025.1031.5
SpatialMLLM-4BParameters=4B2025.1030.5
LLaVA-NeXT-Video-7BParameters=7B2025.1029.9
VG LLM-4BParameters=4B2025.1029.3
LLAVA-Onevision-7BParameters=7B2025.1029