Gemini Omni 1.2 Flash

Model Specs

Item	Specification
Official model ID	gemini-omni-1.2-flash
Core input	Text, images, and video; audio references require allowlisting
Core output	Video with synchronized audio; text responses supported
Generation and editing duration	3–15 seconds in one-second increments
Mode-specific duration	The official model page lists 3–10 seconds for first-and-last-frame output and newly generated extension segments
Output resolutions	360p, 720p, 1080p, and 4K
Aspect ratios	16:9 and 9:16
Sound generation	Speech, music, and sound effects

Overview

Gemini Omni 1.2 Flash became generally available on November 5, 2026. Google positions it for video generation, editing, extension, and upscaling, combining instructions with reference material and follow-up conversational revisions.

Compared with Omni 1.1, Google highlights longer clips, better reference adherence, improved consistency, and more natural dialogue. It supports workflows built around product photography, character references, and scene assets; gated audio-reference input is distinct from generating sound within a video.

Core Capabilities

• Text- and image-to-video: Develop moving scenes from descriptions or an opening image.
• Reference guidance: Convey creative requirements through character, product, style, or short video references.
• Conversational editing: Revise backgrounds, lighting, subjects, or camera direction.
• First-and-last-frame control: Guide transitions with starting and ending images.
• Video extension: Generate continuation segments from existing content.
• Synchronized sound: Generate dialogue, music, and effects for video.
• Video upscaling: Increase the output resolution of selected clips.

Technical Strengths

Feature	Value
Generation and editing up to 15 seconds	Provides room for a short narrative or product explanation.
Multi-turn revisions	Supports continued creative adjustments to existing clips.
Improved reference adherence	Helps preserve character, product, and brand details.
Image and video references	Communicates appearance and movement through visual assets.
Joint audiovisual generation	Incorporates sound design into video creation.
Multiple resolutions and upscaling	Supports reviewing drafts before preparing higher-resolution deliverables.

FAQ

What video workflows suit Gemini Omni 1.2 Flash?

It suits short-video workflows involving references, sound, and iterative revision. Evaluate a real project for subject consistency, motion quality, and editing success rather than judging isolated frames alone.

Does every Gemini Omni 1.2 Flash mode support 15 seconds?

The 15-second limit does not apply uniformly to every mode. The official model page lists 3–10 seconds for first-and-last-frame output and newly generated extension segments, so check the selected mode before configuring a task.

How can I preserve product details with Gemini Omni 1.2 Flash?

Use clear, consistent reference images and identify the colors, shape, packaging, and text that must remain intact. Validate appearance before adding more motion or camera complexity, then review labels and small structures before delivery.

How should I describe conversational edits in Gemini Omni 1.2 Flash?

Specify one main change per turn and state what should remain intact. For example, change the background while retaining the subject’s performance and framing, then review the full clip rather than only the edited area.

How do I use the Gemini Omni 1.2 Flash API through LinkModel?

Open its LinkModel page and follow the published model identifier, authentication method, and task format. Start with a simple clip, then configure references, duration, aspect ratio, and editing modes according to the documentation.

How do I validate Gemini Omni 1.2 Flash on LinkModel?

Use consistent assets to check subject appearance, motion continuity, audio synchronization, and instruction adherence. Record generation duration, task status, and result retrieval, confirming parameter behavior against LinkModel’s documentation.