Longer videos
OpenAI said Higgsfield's video platform can now produce videos longer than 10 minutes in a single run with the GPT-6 Astra model, up from the previous 1-minute limit. That marks a significant jump in how long AI-generated video can run.
Visuals and voice
Two more models are highlighted in the announcement: GPT-Image-2.5 is used for more precise visuals, while GPT-Live is described as a path to voice filmmaking, aiming to add speech and audio layers to video.
What it means
A platform like Higgsfield using these models points to AI video tools moving into longer-form work for ads, short films and content production. OpenAI did not share pricing, regional availability or a release date in the announcement.
