Gemini Omni 1.1 Flash

Overview

Gemini Omni 1.1 Flash is the first generally available Gemini Omni model in the Gemini API. Released August 27, 2026 as gemini-omni-1.1-flash, it turns the June public preview into a stable developer release and adds controls that matter for production creative workflows: scene extension, first/last-frame interpolation, short video references, faster 360p previews, and higher-resolution delivery options.

The model still centers on short video with audio generated from text and multimodal references. Version 1.1 is less about replacing that concept and more about making it controllable and economical across the full iteration loop—from quick composition tests to an upscaled 4K deliverable.

What's New

Scene extension with more context

Omni 1.1 can extend an existing result in 10-second increments up to a 40-second sequence. Google says the model can now analyze up to ten seconds of prior context, compared with previous models that referenced only the final second. That broader temporal window should help preserve action and narrative continuity, although creators still need to check identity, objects, and camera motion at every join.

First- and last-frame interpolation

Developers can provide a starting image and an ending image, then ask the model to generate the transition. This gives art directors stronger boundary control than an open-ended prompt and is useful for product reveals, scene changes, transformation shots, and storyboard-to-video workflows.

Video references

Omni 1.1 accepts short video references of up to three seconds to guide motion, composition, or continuity. Combined with image references, this makes it easier to specify both what a subject should look like and how it should move.

Preview and delivery resolutions

The new 360p mode is intended for fast iteration. Google reports it can be up to 60% faster and cost roughly one-third as much as 720p. Final output options include 720p plus upscaled 1080p and 4K, allowing teams to separate creative selection from delivery rendering.

Performance & Cost Benchmarks

Google publishes vendor-run comparative evaluations for Gemini Omni Flash across video editing, text-to-video, fast motion, image-to-video, and reference-to-video, alongside the workflow measurements and token rates below. These evaluations are not independent tests and are not presented as a version-isolated Omni 1.1 leaderboard.

These are vendor-published operational figures, not independent quality tests. Actual job cost includes inputs, text output, failed generations, and every conversational revision.

Compared With Previous Version

The June 30 gemini-omni-flash-preview model established the API workflow with 3–10 second 720p generation and conversational editing. Version 1.1 keeps that foundation but changes the deployment decision materially.

The most urgent difference is lifecycle: the old preview has a fixed shutdown date, so active integrations need to migrate even if they do not immediately use the new creative controls.

Compatibility Notes

• Replace gemini-omni-flash-preview with gemini-omni-1.1-flash and complete migration before September 30, 2026.
• Re-run prompt and reference regression tests; stable does not guarantee identical framing, motion, or edit behavior.
• The model accepts text, image, and video inputs. Uploaded videos for editing or extension are limited to 10 seconds unless extending model-generated video through multi-turn; video references support up to three clips of up to 3 seconds each. Editing or extending uploaded videos is currently unavailable in the EEA, Switzerland, and the UK.
• Generated output is 3–10 seconds per request at 24 FPS before using extension workflows.
• 1080p and 4K are upscaled outputs, so they improve delivery resolution but do not guarantee additional native scene detail.
• The API's published Gemini Omni 1.1 pricing is paid-tier only; consumer subscriptions and YouTube access follow different quotas and terms.
• All generated output includes SynthID provenance marking.

Pricing & Plans

There is no free Gemini API tier for this model. Use 360p for composition and motion tests, then request higher resolution only for selected clips. Budget for retries and edits rather than multiplying duration by the headline per-second figure alone.

Who Should Upgrade / Who Should Wait

Upgrade now if:

• Your application uses gemini-omni-flash-preview; the September 30 shutdown makes migration time-sensitive.
• You need longer continuous scenes, explicit start/end frames, or motion guidance from a short clip.
• Your creative workflow can save time and money by separating 360p preview from final rendering.
• You need a stable API lifecycle rather than a public-preview dependency.

Wait or evaluate longer if:

• You require native 1080p or 4K generation rather than upscaling.
• Your work depends on perfect character consistency, complex physical interactions, or accurate rendered text.
• A mandatory SynthID watermark conflicts with the distribution requirement.
• You only need occasional consumer creation and do not need the API's programmatic controls.

Sources

• Gemini API changelog
• Google developer announcement
• Gemini API Omni guide
• Gemini Omni model page
• Gemini API pricing
• Gemini API deprecations