Higgsfield uses GPT-6 Astra to extend AI video from 1 minute to over 10 minutes

Higgsfield announced three updates on OpenAI's official channels: using GPT-6 Astra to extend AI video from 1 minute to over 10 minutes; using GPT-Image-2.5 to improve visual precision; and offering a path to voice-driven film production via GPT-Live. The original post did not disclose specific version numbers, pricing, resolution, or availability scope. For AI video creators and short-video teams, breaking the 1-minute single-clip limit means generating fuller narrative segments in one pass, reducing style jumps from repeated stitching; improved visual precision also makes detail-sensitive use cases like brand ads and product demos easier to land. The voice-driven film production path offers a new way to connect dubbing- and narration-led workflows. Note that these capabilities are described only in official messaging; actual usability, API access, and limits are not covered in the original text, so teams relying on this platform should confirm whether their account can use them before adjusting production workflows.