One signal

Update

What changed

OpenAI’s account of Higgsfield’s work adds detail on how its AI tools are being used beyond making longer videos. Higgsfield CEO Alex Mashrabov says creative agents can produce shot variations and use external tools for complex work such as 3D modelling, helping tackle consistency across longer videos.

Mashrabov also says GPT-Image-2.5 can create promotional images with readable text and infographics, a practical boon for small businesses that need more than a pretty picture. He describes Codex as helping creative staff turn ideas into software prototypes, with GPT-6 Astra reducing the engineering work needed to bring those ideas into the product.

Looking ahead, Mashrabov describes a “voice to movie” ambition: creators would speak an idea and see it take shape in real time. That is a stated direction, not a description of a finished feature.

Sources and evidence

• Higgsfield takes AI video from 1 minute to over 10 – OpenAI: OpenAI’s account says Higgsfield’s creative agents can make shot variations and use external tools for 3D modelling; Mashrabov says GPT-Image-2.5 produces images with readable text and infographics, Codex helps creatives prototype software, and the company is aiming towards voice-to-video creation.

Independent WittyWires Watcher; not an official account or feed.