List your company
ResearchFormerly WizWand

Short-answer Visual Question Answering on SEEDBench2+

68.7AccuracyLLaDA-V
40.147.52554.9562.375Apr 8, 2026
Updated 6mo ago

Evaluation results

MethodLinks
LLaDA-VModel Category=Diffusion Vision-Language Models2026.0468.7
Qwen2.5-VL-3BModel Category=Autoregressive Vision-Language Models2026.0468.6
Fast-dVLMModel Category=Diffusion Vision-Language Models, Decoding Strategy=masked diffusion model decoding (MDM)2026.0467.2
Fast-dVLMModel Category=Diffusion Vision-Language Models, Decoding Strategy=speculative decoding (spec.)2026.0467.2
Intern-VL-2.5-4BModel Category=Autoregressive Vision-Language Models2026.0467
LaViDaModel Category=Diffusion Vision-Language Models2026.0457.7
MiniCPM-V-2 (3B)Model Category=Autoregressive Vision-Language Models2026.0452.5
DimpleModel Category=Diffusion Vision-Language Models2026.0451.7
VILA-1.5-3BModel Category=Autoregressive Vision-Language Models2026.0441.2