minimax-h3
MiniMax H3's open license excludes the US, UK and EU
Builds keep appearing under it, including one from NVIDIA Research, while a rival open video model ships with no territory clause at all.
Explainers
How it works and what it means.
Vincent Soberano
Secret Apartment shows how a 60-minute AI film was built
Filipino filmmaker Vincent Soberano won the festival's first Bold Vision Award and its Gold Award for Best AI Major Motion Picture, from more than 1,000 entries.
FlashVSR
FlashVSR+ runs untiled 4K video upscaling on one RTX 4090
The open-source model from Shanghai AI Lab skips the spatial tiling that leaves seams in 4K clips, reporting up to 2x DiT speedup and 45% lower peak memory.
ElevenLabs
ElevenLabs Instant Soundtracks adds music and sound effects
The ElevenCreative update writes original music for an uploaded clip, then times sound effects to moments on screen and places them on the timeline.
fal
fal Worlds streams walkable AI worlds for $4.50 an hour
Lingbot World 2 and Abot World 0 both stream over WebRTC, and the faster one is research only while the Apache-licensed one is cleared for commercial work.
comfyui
ComfyUI's NVIDIA VFX nodes upscale AI video 4x for free
Three nodes that upscale, interpolate and convert to HDR now run in a ComfyUI graph, and frame generation needs a 40-series card.
Claude Code
Claude Code made a 74-shot AI music video with one human
The maker published the storyboard and all 148 prompts for the Suno song, Kling motion and Wan lip-sync as a free GitHub kit.
Theoretically Media
ByteDance's world model re-shoots a movie scene, and Krea 3 lands
Every segment of Theoretically Media's latest episode, checked against the sources it links: a nameless ByteDance world model turning one drawn camera path into a dozen angles inside a Stephen Chow film, Pico's delayed Vision Pro rival, Krea 3's eight-second edits, the report that Sora comes back through a filmmaker startup, an OpenAI engineer charged over a hidden recorder, and two free tools.
wavespeedai
AI Cinema Studio turns 18 camera moves into a shot
WaveSpeedAI's new panel writes lighting, lens and a start-to-end camera move into the prompt, while Kling, Runway, Veo and Luma each specify motion differently.
AI disclosure
How to label AI video on YouTube, TikTok, Instagram and X
Each platform asks whether a viewer could mistake the clip for something real, and the EU's Article 50 adds a legal labeling duty from 2 August 2026.
Grimes
Grimes' Accelerando video: where AI sits in a 70-person pipeline
Grimes' spokesperson says no AI was used for the song or vocals, while the video credits more than 70 professionals over two to three months, including three AI artists.
Kling
AI video's benchmark shifts to holding a character across cuts
Vendors from Kuaishou to Google now sell reference images, saved characters and keyframes as the inputs that keep a person the same across cuts.
animate
animate skill draws explainer videos in code, no video model
The MIT-licensed plugin draws every frame on one canvas, renders with Playwright Chromium and ffmpeg, and asks you to approve the storyboard before it animates.
Code Video Model
Microsoft's Code Video Model plans shots in code first
The research system writes each shot as a Three.js program, previews it as a low-poly proxy, then has the frozen MiniMax-H3 model render the final clip without retraining.
Claude Opus 5.5
How Claude Opus 5.5 painted a hand-drawn music video in code
X user NotinReality handed the lyrics and audio to Claude Opus 5.5, which wrote the JavaScript that paints all 156.6 seconds frame by frame.
youtube
YouTube adds AI music as a fourth disclosure trigger
The help page first showed music in YouTube's core disclosure list on October 3, 2026, a day after it sat only among the examples.
WorldGuide
MBZUAI's WorldGuide makes AI video plan its own next steps
The UAE university's model plans each step from the clip it just generated, scoring 33.33 percent task success against 29.90 percent for MiniMax-H3.
Motionflare
Motionflare Film mode makes a TV-style commercial from a URL
The maker says it writes the brief, plans the shots and renders believable people in real places, with voiceover and sound built into the picture.
California
AI ad performers now need a label in New York, California next
California's law takes effect January 1, 2027 and covers AI voices; New York's has been in force since June 9 and exempts audio-only ads.
Kling
Kling 4.0: what ships today and what is still a promise
Kuaishou's next video model was announced on September 28, but only a lighter Flash tier is usable while the full model waits on a date.