List your company

Video Spatial Understanding on VSI-Bench 500 QA 1.0

49.7AccuracyCoCoSI
34.82838.68942.5546.411Jun 9, 2026

Evaluation results

MethodLinks
CoCoSIBackbone=Gemini-based, Training-free=true, Agentic Collaboration=true, CogMap=true2026.0649.7
CoCoSIBackbone=Gemini-2.5-Flash, Training-free=true, Agentic Collaboration=true, CogMap=true2026.0649.7
Single-Agent BaselineBackbone=Gemini-based, Training-free=true, Agentic Collaboration=false, CogMap=true2026.0644.9
Single-Agent BaselineBackbone=Gemini-2.5-Flash, Training-free=true, Agentic Collaboration=false, CogMap=true2026.0644.9
TraveLERBackbone=Gemini-based, Training-free=true, Agentic Collaboration=true, CogMap=false2026.0644.1
TraveLERBackbone=Gemini-2.5-Flash, Training-free=true, Agentic Collaboration=true, CogMap=false2026.0644.1
VCABackbone=Gemini-based, Training-free=true, Agentic Collaboration=false, CogMap=false2026.0643.6
VCABackbone=Gemini-2.5-Flash, Training-free=true, Agentic Collaboration=false, CogMap=false2026.0643.6
VideoAgentBackbone=Gemini-based, Training-free=true, Agentic Collaboration=false, CogMap=false2026.0643.3
VideoAgentBackbone=Gemini-2.5-Flash, Training-free=true, Agentic Collaboration=false, CogMap=false2026.0643.3
CoCoSIBackbone=GPT-based, Training-free=true, Agentic Collaboration=true, CogMap=true2026.0643
CoCoSIBackbone=Qwen-VL-based, Training-free=true, Agentic Collaboration=true, CogMap=true2026.0641.1
CoCoSIBackbone=Qwen2.5-VL-32B, Training-free=true, Agentic Collaboration=true, CogMap=true2026.0641.1
Single-Agent BaselineBackbone=GPT-based, Training-free=true, Agentic Collaboration=false, CogMap=true2026.0639.7
TraveLERBackbone=GPT-based, Training-free=true, Agentic Collaboration=true, CogMap=false2026.0639.3
VCABackbone=GPT-based, Training-free=true, Agentic Collaboration=false, CogMap=false2026.0638.9
VideoAgentBackbone=GPT-based, Training-free=true, Agentic Collaboration=false, CogMap=false2026.0638.4
Single-Agent BaselineBackbone=Qwen-VL-based, Training-free=true, Agentic Collaboration=false, CogMap=true2026.0636.6
Single-Agent BaselineBackbone=Qwen2.5-VL-32B, Training-free=true, Agentic Collaboration=false, CogMap=true2026.0636.6
TraveLERBackbone=Qwen-VL-based, Training-free=true, Agentic Collaboration=true, CogMap=false2026.0636.2
TraveLERBackbone=Qwen2.5-VL-32B, Training-free=true, Agentic Collaboration=true, CogMap=false2026.0636.2
VCABackbone=Qwen-VL-based, Training-free=true, Agentic Collaboration=false, CogMap=false2026.0635.7
VCABackbone=Qwen2.5-VL-32B, Training-free=true, Agentic Collaboration=false, CogMap=false2026.0635.7
VideoAgentBackbone=Qwen-VL-based, Training-free=true, Agentic Collaboration=false, CogMap=false2026.0635.4
VideoAgentBackbone=Qwen2.5-VL-32B, Training-free=true, Agentic Collaboration=false, CogMap=false2026.0635.4