List your company

Video Understanding on Tomato (test)

0.53AccuracyGPT-5
0.234640.311320.3880.46468Jan 15, 2026Jan 27, 2026Feb 8, 2026Feb 21, 2026Mar 5, 2026Mar 17, 2026Mar 30, 2026

Evaluation results

MethodLinks
GPT-5Access Type=API call only2026.010.53—
GPT-5Category=API call only2026.030.53—
Gemini 2.5 ProAccess Type=API call only2026.010.486—
Gemini 2.5 ProCategory=API call only2026.030.486—
Gemini 3 ProAccess Type=API call only2026.010.483—
Gemini 3 ProCategory=API call only2026.030.483—
GPT-5 miniAccess Type=API call only2026.010.441—
GPT-5 miniCategory=API call only2026.030.441—
Molmo2-4BAccess Type=Molmo2 family2026.010.398—
Molmo2-4BCategory=Fully open2026.030.398—
Claude Sonnet 4.5Access Type=API call only2026.010.396—
Molmo2-8BAccess Type=Molmo2 family2026.010.396—
Claude Sonnet 4.5Category=API call only2026.030.396—
Molmo2-8BCategory=Fully open2026.030.396—
Gemini 2.5 FlashAccess Type=API call only2026.010.391—
Gemini 2.5 FlashCategory=API call only2026.030.391—
MolmoPoint-8BCategory=MolmoPoint2026.030.373—
Molmo2-O-7BAccess Type=Molmo2 family2026.010.362—
Molmo2-O-7BCategory=Fully open2026.030.362—
Qwen3-VL-8BAccess Type=Open weights only2026.010.357—
Qwen3-VL-8BCategory=Open weights2026.030.357—
PLM-8BAccess Type=Open models2026.010.332—
PLM-8BCategory=Fully open2026.030.332—
Keye-VL-1.5-8BAccess Type=Open weights only2026.010.33—
Keye-VL-1.5-8BCategory=Open weights2026.030.33—
VideoChat-Flash-7BAccess Type=Open models2026.010.325—
VideoChat-Flash-7BCategory=Fully open2026.030.325—
Qwen3-VL-4BAccess Type=Open weights only2026.010.318—
Qwen3-VL-4BCategory=Open weights2026.030.318—
Eagle2.5-8BAccess Type=Open weights only2026.010.31—
Eagle2.5-8BCategory=Open weights2026.030.31—
PLM-3BAccess Type=Open models2026.010.309—
PLM-3BCategory=Fully open2026.030.309—
GLM-4.1V-9BAccess Type=Open weights only2026.010.3—
GLM-4.1V-9BCategory=Open weights2026.030.3—
MiniCPM-V-4.5-8BAccess Type=Open weights only2026.010.298—
MiniCPM-V-4.5-8BCategory=Open weights2026.030.298—
InternVL3.5-4BAccess Type=Open weights only2026.010.268—
InternVL3.5-4BCategory=Open weights2026.030.268—
LLaVA-Video-7BAccess Type=Open models2026.010.249—
LLaVA-Video-7BCategory=Fully open2026.030.249—
InternVL3.5-8BAccess Type=Open weights only2026.010.246—
InternVL3.5-8BCategory=Open weights2026.030.246—
BaselineModel=Baseline, Pruning Strength=0%2026.03—0.365
InternVL3.5-8BModel=InternVL3.5-8B, Pruning Strength=None2026.03—0.246
PLM-8BModel=PLM-8B, Pruning Strength=None2026.03—0.332
Qwen3-VL-4BModel=Qwen3-VL-4B, Pruning Strength=None2026.03—0.318
STTSModel=STTS, Pruning Strength=30%2026.03—0.356
STTSModel=STTS, Pruning Strength=40%2026.03—0.346
STTSModel=STTS, Pruning Strength=50%2026.03—0.351