Generate images from text prompts
VGGT (CVPR 2025)
OpenAI's Deep Research, but open
Estimate human poses in images