Flux 3
Black Forest Labs says FLUX 3 is its first unified model trained across images, video, and audio for generation and action prediction.
The model is in early access for video, with native audio generation and clips up to 20 seconds. The company says early evaluations show FLUX 3 beating several rival video models in preference tests, while noting the results are preliminary. Image generation, action prediction with partners, private weight access, and an open-weight multimodal backbone are planned after staged early access. HN · Frontpage AI's note
The model is in early access for video, with native audio generation and clips up to 20 seconds. The company says early evaluations show FLUX 3 beating several rival video models in preference tests, while noting the results are preliminary. Image generation, action prediction with partners, private weight access, and an open-weight multimodal backbone are planned after staged early access. HN · Frontpage AI's note
score 8
Discussions
- hn · 284 points · 67 comments
- hn · 313 points · 77 comments
- hn · 336 points · 86 comments
- hn · 361 points · 92 comments
- hn · 387 points · 96 comments
- hn · 407 points · 101 comments
- hn · 424 points · 107 comments
- hn · 438 points · 111 comments
- hn · 453 points · 112 comments
- hn · 469 points · 115 comments
- hn · 484 points · 115 comments
- hn · 496 points · 118 comments
- hn · 506 points · 121 comments
- hn · 510 points · 121 comments
- hn · 516 points · 124 comments
- hn · 520 points · 125 comments
- hn · 531 points · 125 comments
- hn · 535 points · 125 comments
- hn · 537 points · 125 comments
- hn · 540 points · 127 comments
- hn · 543 points · 128 comments