TBC
TBC plans to sell its neuron-taught video model through AWS
The San Francisco startup says its software layer, measured from living neural activity, speeds up text-to-video inference fivefold, though it will not name the base model.
News
Launches, updates, demos and everything that shipped.
Runway
Runway Labs demos video that redesigns itself to fit the frame
A seven-post thread on a 1,310-follower account shows generated video reflowing to the window, and a straight staircase redrawn as a spiral one.
Higgsfield
Higgsfield says one engineer shipped video features in a day
In OpenAI's customer story, Higgsfield's chief executive credits the model's long-horizon task planning for video ad features that now need one engineer.
HelloGen
HelloGen sells unlimited AI video on one model for $19.90
The service removed per-clip charging on MiniMax H3 and published a floor instead, with Pro subscribers guaranteed two priority renders a day and Max five.
MVLAND
MVLAND 2.0 adds shot-level editing to AI music videos
Meitu's music video generator now analyses a track before it generates anything, then lets creators re-do the camera, composition or action of a single scene.
Meshy
Meshy's Mora builds a world in 3D before rendering it
The 3D generation company's research architecture hands game logic, spatial structure and final imagery to three different layers, and a playable demo is live.
AI likeness
Zelda Williams asks fans to stop making AI videos of her father
The director says a video circulating as private footage of Robin Williams is "clearly AI", nearly a year after she first asked fans to stop recreating him.
OpenArt
OpenArt Arena has 1,031 creatives rank AI image and video models
OpenArt Arena ranks models by blind judging across job categories, and the company's own disclosures page states what the rankings do not establish.
Meta
Pin app blames a Meta ad tool for lightening a creator's skin
The app says Advantage+ Creative Standard Enhancements were switched on by default and generated derivatives of its ads, and that it missed an $85,000 pre-suit demand deadline.
iQIYI
iQIYI contest requires 60% of each AI film to come from Wan 3.0
Submissions close September 23 for the contest on NadouPro, which requires at least 60% of each entry to come from Alibaba's Wan3.0 and 80% to be made on the platform.
freebeat
freebeat Launches a Lyric Video Maker That Builds the Words Into the Scene
The San Francisco company says its new tool generates a typographic treatment for each scene and matches it to the track's tempo, energy curve and camera motion.
duel
Duel Apologizes for an AI Video of Sneako Praying on Its Blackjack Table
The crypto casino says its own content team made the clip and that an employee was fired; the streamer says he will sue, and no case has been filed publicly.
creatify labs
Creatify Labs' Boreal Sells AI Video at 1 Cent a Second, Built for Ads
The advertising-focused model was post-trained on an open 22-billion-parameter base and returns a five-second 720p clip for about $0.05, the company says.
astronex
Astronex-World 1.0 Streams Open World-Model Video on a Single 48GB GPU
The Apache-2.0 release is built on Alibaba's Wan 2.2 prior and generates 832x480 video at 24 frames per second without a render queue, its authors say.
Fountain 0
Fountain 0 Sells a Feature-Length All-AI Odyssey Film for $9.99
The studio behind Tribeca's Dreams of Violets says its second feature ran on Kling and took months of part-time work, and reviewers have not been kind.
Alibaba
Wan 3.0 generates 30 seconds of video in one pass
Alibaba's Wan 3.0 doubles its predecessor's single-pass clip length to 30 seconds, but its per-second price beats or trails Wan 2.7 depending on which platform bills it.
Kling 3.0
Kling 3.0's 4K 60fps claim traces to a single press release
Kuaishou confirmed 15-second multi-shot clips and five-language native audio for Kling 3.0; the widely repeated 4K 60fps and six-shot figures trace to a single marketing release.
HiDream
HiDream.ai's HiDream V1 Video Model Debuts at 1080p With Synced Audio
HiDream.ai's new omnimodal model can pick its own clip length between 5 and 20 seconds, and the company says it debuted at No. 4 and No. 8 on two video leaderboards.
Vidu
ShengShu's Vidu S2 Runs Interactive 720p Avatars and Live Video Edits
The follow-up to Vidu S1 lifts real-time avatar output from 540p to 720p and restyles an incoming camera feed as it plays.
Tilly Norwood
AI actress Tilly Norwood switches to Chinese on Piers Morgan
The glitch came during a press tour for Misaligned that, according to TechCrunch, put the AI character in front of 75 journalists at once.