Skip to content
WattsUpNext
Video editing timeline with blue and purple clips displayed on dark screen

AI

Higgsfield says AI videos now run past 10 minutes, up from one

Higgsfield's CEO says GPT-6 Astra lets users make AI videos longer than 10 minutes, where the limit used to be about one. The claim comes from OpenAI's own customer story.

Photo by Peter Stumpf on Unsplash · Unsplash License · source

By WattsUpNext Desk · Edited by Juda B. Hur2 min read

Higgsfield AI says its users can now make videos longer than 10 minutes, up from a ceiling of around one minute before OpenAI's GPT-6 Astra model arrived. That's a claim from CEO and co-founder Alex Mashrabov, published on OpenAI's own site, so treat it as a company account rather than an independent test.

If it holds up, it matters to anyone who has watched AI video stay stuck at short clips. Ten minutes is the difference between a social teaser and something closer to a short film or a full lesson.

How Higgsfield says it got there

The hard part of long AI video is consistency: keeping a character, outfit or set looking the same from shot to shot. Mashrabov says Higgsfield's creative agents, which handle repetitive jobs such as producing variations of a shot, can now call outside tools with GPT-6 Astra. That includes tools for complex 3D modeling work, which he says helps with consistency across longer videos.

He also says the model and GPT-Image-2.5 let a single creator run a high-end fashion campaign from start to finish. He called GPT-Image-2.5 "state of the art" and said it is the first of its kind to reliably render text and infographics on top of imagery. Those are his words, and the post offers no benchmarks or sample videos.

Codex, classrooms and what's still a plan

Higgsfield says it has about 200 creatives and 100 engineers working closely together. Creatives use OpenAI's Codex to turn ideas into software prototypes, and Mashrabov says GPT-6 Astra cuts the work needed to move those into the product. He also described an early effort to use the model's visual understanding to spot topics that rural students in Kazakhstan struggle with, then make tailored instructional videos.

The next goal is what he calls "voice to movie": you describe an idea out loud and watch it take shape in real time. He praised OpenAI's voice model, GPT-Live, as "best in class for voice understanding." That is an ambition, not a shipped feature, and the post gives no date, pricing or quality details for long videos.

What to watch: whether Higgsfield shows a full 10-minute video that stays consistent from start to finish, since the post shows none. Would you watch a feature made that way?

Sources

  1. 1.Higgsfield takes AI video from 1 minute to over 10 · OpenAI

Reported by the WattsUpNext desk from the sources linked below. Spot an error? Tell us at corrections@wattsupnext.com.

The WattsUpNext Brief

Get stories like this in your inbox.

One email each weekday, only the topics you choose. Real news, sourced, no fluff. Unsubscribe in one click.

What would you like to get?

Keep reading

Published Oct 14, 2026