Higgsfield AI says OpenAI’s newer models let its users create AI videos longer than ten minutes, up from a ceiling of about one minute before GPT-6 Astra. CEO Alex Mashrabov describes the change as enabling smaller creative teams to take on bigger projects, including a single creator executing high-end fashion campaigns from start to finish. Higgsfield’s creative agents handle repetitive production work—for example, generating multiple variations of a shot—and with GPT-6 Astra they can call external tools for manual 3D modeling tasks. Mashrabov ties that capability to one of creative AI’s hardest problems: maintaining consistency across longer videos.

The company’s own YouTube channel, close to half a million subscribers, shows the practical result: instructional videos over ten minutes are now produced with a GPT-6 Astra-based workflow instead of requiring considerable manual effort. Mashrabov also relates the longer-video capability to a personal project supporting rural schools in Kazakhstan, where GPT-6 Astra‘s visual understanding can help identify topics students find difficult and generate learning videos tailored to their needs.

For small businesses creating promotional content, GPT-Image-2.5 is presented as a step forward for readable text and precise infographics. Mashrabov calls it ‘state of the art and the first model of its kind that can reliably render text and infographics on top of the imagery,’ allowing businesses without large production budgets to create materials where text accuracy matters as much as the image.

Higgsfield AI builds its product with close collaboration between about 200 creatives and 100 engineers. Codex gives the creative team ‘the power of a really strong software engineer,’ letting them turn ideas into software prototypes. In the past, bringing those prototypes into production required heavy front-end and back-end engineering involvement; now, Mashrabov says GPT-6 Astra reduces that work and can ship production-ready code, which he says was never possible before. The result is less friction between prototyping and production, and better products shipped faster.

Looking ahead to what the article frames as voice-to-video with GPT-Live-1, Higgsfield wants creators to describe an idea and see it take shape as they talk. Mashrabov sees OpenAI’s advances in planning, tool calling, image generation, and voice bringing that experience closer, and he describes OpenAI’s voice model as ‘best in class for voice understanding and for regular communication.’ The stated ambition is to make creating a film feel more like describing an idea, with AI handling more of the work in between.