List your company

Task success rate on DMV-Bench J=5 1.0

82.7Task Success RateDualMem
-3.30819.02141.3563.679Jun 25, 2026

Evaluation results

MethodLinks
DualMemBack-end=Gemini 2.5 Flash, Steps (stp)=118–134, Probe counts (nr)=2,7622026.0682.7
DualMemBack-end=Qwen2.5-VL-7B, Steps (stp)=118–134, Probe counts (nr)=1,0532026.0681.1
M2ABack-end=Qwen2.5-VL-7B, Steps (stp)=118–134, Probe counts (nr)=1,0532026.0670.4
CaptionBack-end=Qwen2.5-VL-7B, Steps (stp)=118–134, Probe counts (nr)=1,0532026.0667.3
M2ABack-end=Gemini 2.5 Flash, Steps (stp)=118–134, Probe counts (nr)=2,7622026.0665.7
CaptionBack-end=Gemini 2.5 Flash, Steps (stp)=118–134, Probe counts (nr)=2,7622026.0658.9
MMABack-end=Qwen2.5-VL-7B, Steps (stp)=118–134, Probe counts (nr)=1,0532026.0647.7
MMABack-end=Gemini 2.5 Flash, Steps (stp)=118–134, Probe counts (nr)=2,7622026.0646.1
WorldMMBack-end=Gemini 2.5 Flash, Steps (stp)=118–134, Probe counts (nr)=2,7622026.0643.5
WorldMMBack-end=Qwen2.5-VL-7B, Steps (stp)=118–134, Probe counts (nr)=1,0532026.0639.8
NoMemoryBack-end=Qwen2.5-VL-7B, Steps (stp)=118–134, Probe counts (nr)=1,0532026.060
TextOnlyBack-end=Qwen2.5-VL-7B, Steps (stp)=118–134, Probe counts (nr)=1,0532026.060
NoMemoryBack-end=Gemini 2.5 Flash, Steps (stp)=118–134, Probe counts (nr)=2,7622026.060
TextOnlyBack-end=Gemini 2.5 Flash, Steps (stp)=118–134, Probe counts (nr)=2,7622026.060