# Andreas Blattmann

Markdown copy of https://realalpha.app/people/andreas-blattmann/ on Real Alpha, the best minutes from long interviews with the people building AI. Cite and link the page, not this copy.

Real Alpha has cut 4 moments of Andreas Blattmann (Black Forest Labs) talking, from an interview on Latent Space published Nov 16, 2026.

- Role: Black Forest Labs

## What Andreas Blattmann said

- [A video model learned enough 3D to fine-tune in 1,000 steps](https://realalpha.app/s/a-video-model-learned-enough-3d-to-fine-tune-in-1-000-steps/)
  Latent Space · Nov 16, 2026 · 3 min
  > “I think, 1K fine-tuning steps on 3D data, and we got a 3D model out of it.” — Andreas Blattmann · [36:51](https://realalpha.app/v/8cGfJOU3AlQ/?t=2211)
- [Training should focus compute on informative denoising steps](https://realalpha.app/s/training-should-focus-compute-on-informative-denoising-steps/)
  Latent Space · Nov 16, 2026 · 5 min
  > “where should we spend our compute? Which parts of this trajectory from data to noise carry the information that's most important?” — Andreas Blattmann · [44:33](https://realalpha.app/v/8cGfJOU3AlQ/?t=2673)
- [BFL aims to crack one-shot robot learning by 2027](https://realalpha.app/s/bfl-aims-to-crack-one-shot-robot-learning-by-2027/)
  Latent Space · Nov 16, 2026 · 8 min
  > “You should be showing a robot an action one time, and then it should be able to do this” — Andreas Blattmann · [1:25:29](https://realalpha.app/v/8cGfJOU3AlQ/?t=5129)
- [Autoregression is more data-efficient; diffusion can sample faster](https://realalpha.app/s/autoregression-is-more-data-efficient-diffusion-can-sample/)
  Latent Space · Nov 16, 2026 · 5 min
  > “I would say autoregressive models are way more data efficient.” — Andreas Blattmann · [1:14:14](https://realalpha.app/v/8cGfJOU3AlQ/?t=4454)

## The episode

- [Why FLUX 3 Trains on Images, Video and Sound — Andreas Blattmann & Dustin Podell, Black Forest Labs](https://realalpha.app/v/8cGfJOU3AlQ/) · Latent Space · Nov 16, 2026
