List your company

Negative Query Recognition on CoMET-Bench

68.1Rej.-F1CoMET-Agent (Gemini 3 Flash)
-2.72415.66334.0552.437Jun 13, 2026

Evaluation results

MethodLinks
CoMET-Agent (Gemini 3 Flash)#Size=-2026.0668.15.8
CoMET-Agent (GPT-5)#Size=-2026.0666.56.4
InternVL3.5#Size=8B2026.0664.716.8
Gemini 3 Flash#Size=-2026.0661.551.7
VideoARM#Size=-2026.06619.5
Gemini 2.5 Pro#Size=-2026.0660.837.3
GPT-5#Size=-2026.0660.65.3
CoMET-Agent (Qwen3-VL)#Size=8.6B2026.0657.48
Qwen3-VL#Size=8.3B2026.0652.15.3
LLaVA-Video#Size=7B2026.0648.958.5
Qwen3-VL#Size=30B2026.0647.25.9
InternVL3.5#Size=38B2026.0640.39.1
Eagle2.5#Size=8B2026.0639.914.6
DisTime#Size=7B2026.0612.193.5
LLaVA-OV1.5#Size=7B2026.069.32.6
MiMO-VL#Size=7B2026.068.83.8
LongVILA-R1#Size=7B2026.065.697.1
TRACE#Size=7B2026.062.598.8
LITA#Size=13B2026.061.799.1
VideoMind#Size=7B2026.061.299.4
TimeLens#Size=7B2026.060100