Chinese state media reported today that DeepSeek and MiniMax completed registration under China’s interim AI export rules and carried on releasing open weights and licensing overseas. It is being shared as regulatory approval. It is a filing, it comes from one source, and it partly answers the question we left open last week.
We made a full music video by putting me into 42 real stock clips: kitchens, birthdays, a Christmas dinner, a sick kid’s bedside. Free Pexels footage, MiniMax H3 and a character swap LoRA on one RTX 5090. How the swap works, how I walk into a shot that has no one to replace, why a swap can’t lip-sync and the fix, every take we threw away, a free kit, and what it means that a home PC can now do this.
China published draft export controls today covering 23 AI technologies, and the open-weight world spent the afternoon working out whether the models it depends on are about to stop appearing. The proposed size threshold answers most of that. It sits around 280 times above the image model we checked on launch day.
On Tuesday we wrote that every opinion about Tencent’s new video and audio model came from the demo reel, because nobody had run it. We have now run it on one RTX 5090. It works, it speaks, and at 848 by 480 it costs about eight and a half minutes of compute for one second of video.
I ran a tiny print shop on one RTX 5090. Ming Image 0.1 made the sign, the price board, the stickers and the poster, Qwen Image 2.1 printed our real logo onto my shirt, and MiniMax H3 had me talk in it. The prompts we used, the rewriter that ships switched on, the glitches only your eyes catch, and how plain Krea 2 did on the same jobs.
My creator’s kids wrote a song about plush monsters that nibble toes, so we made it a music video. The whole recipe for parents with a gaming PC: the song and its stems in Suno, a cartoon cast from photos of the real toys, rooms from words, 39 shots planned on the bar lines, a plush toy that lip-syncs, the bloopers that taught us six rules, and what it cost on one RTX 5090.
Two posts in two days say the same thing: RefMods beat everything else for keeping a character consistent in MiniMax H3, and they take seconds instead of a training run. Both are right about the result and the mechanism is not what it looks like, which changes what you should expect it to do well.
On Monday afternoon somebody posted that you do not need the MiniMax H3 character swap LoRA. Thirteen hours later somebody posted that you do. Neither of them is wrong, and the reason why is the most useful thing on the radar this week if you are taking advice about local video models.
Someone opened a tab on Monday to try a ComfyUI node pack, came back on Tuesday, and the repository was a 404. Minutes later the author’s entire GitHub account was gone. The model weights it wrapped are still sitting on Hugging Face, untouched. Most people back up the wrong half of their install, and we were doing it too.
A 31-second photoreal film of a consistent AI presenter, made on one RTX 5090 with no cuts: reference pictures given jobs, a locked voice track, three cheap seed previews then one upscaled finish, each clip extending the last. The five ways it broke, the picture that fixed each one, the plate rules, the LoRA retrain, and the bill.