Gemini Omni Flash goes GA: 4K video and clip extension on the API

Google promoted its video model to production — with a Sept 30 preview-endpoint cutoff and image-to-video interpolation — as Copilot's review default flips.

Nowline AUG 31 5:00 AM banner

Top AI stories from the last hour

Top AI stories from the last hour

Copy markdown

  • 4K output and clip extension, now stable

    The production model, `gemini-omni-1.1-flash`, adds a resolution control for 360p, 720p, 1080p and 4K, and can extend an existing clip by generating a continuation at its end. Chain generations to build longer sequences instead of being capped at one short clip.

  • Move off the preview endpoint by Sept 30

    `gemini-omni-flash-preview` sunsets September 30, 2026 — calls to it break after that date. Swap to `gemini-omni-1.1-flash` now; the task types match and you pick up the resolution and extension params for free.

  • Weekend build: tween between two stills

    Interpolation ships as an `image_to_video` task that generates the transition between two images. Feed a before/after pair — a rough sketch and its finished art, a product's open and closed state — and get a smooth animated tween with no keyframing.

  • Elsewhere: Copilot code review defaults to 'Balanced' Sept 28

    GitHub flips the Copilot code-review default from Lite to Balanced across all repos on September 28 — more thorough passes, but they spend more premium requests; set it back to Lite beforehand to keep today's behavior. New Copilot Business and Enterprise seats also must be paid before access (new signups Sept 1, existing orgs Oct 1).