Gemini Omni Flash

Create anything from any input with conversational video editing

Context window

Not announced

Input / 1M tokens

Not announced

Output / 1M tokens

Not announced

Provider

Google

Data verified 2026-07-17

Gemini Omni Flash is Google's first any-to-any multimodal model that accepts text, images, audio, and video as inputs and produces video with synchronized audio. It features conversational, multi-turn editing that preserves character consistency and physics across edits. Currently capped at 10-second clips as a deployment choice.

Capability index

Relative estimates (0-100) to place this model against its peers, grounded in published benchmarks.

Coding

Reasoning

Math

Multimodal

Long context

Speed

Cost efficiency

How to access it

Available in Gemini app, Google Flow, YouTube Shorts, and YouTube Create. Requires Google AI Plus ($7.99/month minimum) for non-YouTube access. Developer/enterprise API access announced but not yet available.

Get access →Documentation →

Strengths

✓True multimodal input handling (text, image, audio, video)
✓Conversational multi-turn editing with context preservation
✓Character and physics consistency across edits
✓Synchronized audio generation in single pass
✓Available free on YouTube Shorts and YouTube Create

Best for developers who...

Content creators needing iterative video editingBusinesses creating marketing and training videosUsers wanting to generate videos with custom avatarsVideo editing without timeline manipulation

When to choose it (and when not to)

Reach for Gemini Omni Flash when...

→When you need step-by-step conversational editing
→When generating videos from mixed media inputs
→When character consistency is critical across edits
→For short-form video creation (up to 10 seconds)

Look elsewhere if...

✕Generating videos longer than 10 seconds
✕Enterprise workflows requiring API access (not yet available)
✕Projects prioritizing maximum video quality over editability

How to use it

›Use natural language descriptions for edits
›Each instruction builds on the previous one in a conversation
›Provide reference images or sketches for better control
›Describe physics-based scenarios for realistic interactions

Quickstart

Web Interface

// Access via Gemini app or at gemini.google.com
// Select Gemini Omni Flash model
// Enter text prompt: "A marble rolling on a wooden track"
// Optionally attach reference images
// Click Generate to create video
// Use chat to edit: "Make it nighttime instead"

Web interface and mobile app; Python SDK for Gemini API coming in weeks

API model id: gemini-omni-flash

Compare Gemini Omni Flash

Compare Gemini Omni Flash with any other model

Build a comparison →All model comparisons →

Learn the concepts

Inference Context Window

← All AI models