descriptive claim
The speaker says that about two to three years before the conversation, most generative modeling for images and video shifted from diffusion models to flow-matching models.
Jason Ma · Dyna-2: General Purpose Physical Agents with Hour-Long Autonomy — Jason Ma, Dyna Robotics · 13 Oct 2026
In their words
The transcript segments this claim was taken from
“That is why I think about two to three years ago,”
“most generative modeling for image and video has shifted from”
“a diffusion model to a full matching model but the concept is the same and full matching is just a special simpler to tune a variant of diffusion yeah yeah”Play from 6:10Report a misattribution
80%how strongly it was put, not whether it is true
70%importance to the conversation
82%confidence the claim was read correctly