Flux 3

FLUX 3 has arrived in Early Access promising a very big dream: one AI system that can understand and create images, video, and sound together. The company says that means more realistic clips, better timing between what you see and hear, and even a step toward machines that understand the real world. It also boasts some early win rates against rival video tools, plus flashy features like 20-second video generation, animation from still images, and built-in audio.

But in the court of public opinion, the comment section became the real main event. The biggest mood? Deep, eye-rolling skepticism. One commenter dragged the launch for showing "close to zero examples of people," mocking the grand talk about a "world model" while asking where the proof is. Another pounced on the branding itself, joking that if the company keeps saying "visual intelligence" while bragging about audio too, maybe nobody proofread the pitch. Ouch.

Then came the classic internet roast: accusations of AI-written marketing fluff. One reader said the opening paragraphs had the "stench of LLM slop writing" and admitted they mentally checked out on the spot. Others got nitpicky in the funniest possible way, like the person asking why the post talks about learning from images, video, and audio separately when, uh, video already contains images and audio. Meanwhile, one practical commenter dug into the fine print and noted that access looks pretty locked down, with private weights and partner-only plans. Translation: cool demo, but many readers are still asking the oldest tech question of all — can regular people actually use it, and does it really do what the hype says?