Higgsfield Unveils GPT‑6Astra, Pushing AI Video Generation to One Minute

Higgsfield has announced the release of its GPT‑6Astra model, which extends the maximum length of AI‑generated videos from the previous 10 seconds to over one minute. The new model supports video creation ranging from 10‑second clips to full‑minute sequences, marking a significant step forward in the temporal scope of generative video content. In addition to the extended duration, Higgsfield introduced GPT‑Image‑2.5, a visual‑enhancement module that improves frame fidelity and overall image quality within the generated videos. The company also unveiled a GPT‑Live pathway, enabling real‑time voice‑driven film production. These developments collectively broaden the creative possibilities for users, allowing for longer, higher‑quality video outputs and more interactive, voice‑controlled storytelling. The announcement follows earlier reports that highlighted Higgsfield’s ambition to push AI video beyond the 10‑second limit, and the new release confirms that ambition with tangible improvements in both length and visual quality. The GPT‑Live feature further differentiates Higgsfield’s offering by providing a streamlined workflow for voice‑based film creation, potentially accelerating production timelines for content creators.

Written with AI from the reporting · updated 10/14/2026