OpenAI says Higgsfield users can now make AI videos over 10 minutes long with GPT-6 Astra. Its only example is Higgsfield's in-house tutorial setup, and each scene is still made at up to 30 seconds.
On Oct 13, OpenAI published a customer story headlined "Higgsfield takes AI video from 1 minute to over 10." It says that before GPT-6 Astra, Higgsfield videos topped out at around one minute, and that users can now create videos over 10 minutes long. Within hours, accounts on X were reporting that Higgsfield had raised its length limit tenfold.
The page itself says something narrower. The 10-minute line is OpenAI's own wording, and no direct quote from CEO Alex Mashrabov mentions minutes. Its only example is a setup Higgsfield built to make instructional videos for its own YouTube channel. GPT-6 Astra is OpenAI's language model: it plans shots and runs other tools, and it does not make video itself. Higgsfield's Cinema Studio still makes each scene at up to 30 seconds, and the channel posted Hell Grind, a 95-minute AI feature, on Aug 4, a month before its first Astra tutorial.
So the honest version is smaller and more interesting. Higgsfield says its AI can now put a long video together from start to finish, a job that used to need people at every step, while OpenAI shows no sample, no price and no customer film made this way.
In practice, a long AI video is still a run of short scenes. Plan longer pieces as 15 to 30 second scenes, try a language model as your shot planner and continuity checker, and leave real time for the edit; Higgsfield's own 11-minute "One Chat" video credits After Effects and DaVinci Resolve.
Source: OpenAI, Higgsfield on YouTube, Oct 13, 2026.
