Gemini Omni Flash: Video Quality, Pricing, and Test Results

👁 121 reads

Updated continuously. Current as of October 9, 2026.

Gemini Omni Flash is Google DeepMind’s video generation model that replaced Veo inside the Gemini app. It produces 3-10 second clips with synchronized audio from text, images, and short videos, extends scenes up to 40 seconds, and supports conversational video editing. On the Artificial Analysis image-to-video leaderboard, it ranks 4th with an Elo score of 1178 - one point behind Vidu Q4 and three behind MiniMax H3. It is available through Google AI consumer plans and runs about $0.10 per second for 720p video via the API, though user-uploaded video editing remains restricted in Europe.

What is Gemini Omni Flash?

Gemini Omni is a family of multimodal models unveiled by Google at I/O on May 19, 2026. According to Google DeepMind, it represents the first step toward a system capable of “generating anything from anything,” starting with video. Omni merges the broad world understanding of Gemini models with the video synthesis capabilities originally developed for Veo. DeepMind’s Nicole Brichtova told TechCrunch that Omni is not simply a Veo update, but a fusion of Gemini intelligence with native generative media architectures.

Flash is the first and currently only released variant. Here is how its rollout unfolded:

• May 19, 2026 - Launched in the Gemini app, Google Flow, and YouTube Shorts.
• June 30, 2026 - Released in preview via the Gemini API (gemini-omni-flash-preview).
• August 27, 2026 - Gemini Omni 1.1 Flash reached general availability, adding scene extension, first-and-last frame prompting, 4K upscaling, and low-cost 360p previews.
• September 30, 2026 - The preview endpoint was deprecated and removed.
• October 22, 2026 - Earliest deprecation date for Veo 3.1 preview in the Gemini API, with Omni serving as its primary successor (covered in our guide on Veo 3.1 alternatives).

Google has also teased Omni Pro, intended to launch once it delivers a decisive generational leap over Flash. No release date has been confirmed.

Key Specifications

How to Access Gemini Omni Flash

1. Gemini App: Requires an active Google AI Plus, Pro, or Ultra subscription (18+ account). Inside the chat window, select video generation, enter a prompt, or attach up to 5 reference images. Free accounts do not have video generation access.
2. Google Flow (labs.google/flow): Select Gemini Omni Flash from the model picker. A 720p clip consumes 7-15 credits depending on length, while 360p costs 4-5 credits. Read our full walkthrough in the Google Flow guide.
3. Gemini API: Obtain an API key via Google AI Studio, select gemini-omni-1.1-flash, and call it via the Interactions API. Omni has no free tier in the API.
4. Third-Party Cloud Providers: Hosted endpoints are available on fal.ai (google/gemini-omni-flash/v1.1/text-to-video) and Replicate (google/gemini-omni-1.1), billed directly per generation. We conducted our benchmarks through fal.ai.

Regional restrictions: Editing and extending user-uploaded videos is disabled across the EEA, Switzerland, and the UK. In those regions, you can only extend and modify clips that the model itself generated. Uploading and editing images containing minors is also blocked in these jurisdictions.

Gemini Omni Flash Pricing

Consumer access inside the Gemini app and Google Flow operates on monthly subscription credits:

Google does not publish hard daily clip limits for the Gemini app - quotas scale based on computational compute multiplied by subscription tier. Through developer APIs, pricing is billed strictly per second:

The 720p rate reflects Gemini API token pricing (5,792 tokens per second at $17.50 per million tokens, working out to ~$0.10), while other resolutions reflect fal.ai’s hosting prices. One minute of 720p video costs roughly $6.00, aligning with Artificial Analysis benchmarks. You can inspect cost comparisons across every leading tool in our breakdown of AI video generation costs per second.

Artificial Analysis Benchmark Rankings

Artificial Analysis ranks models using blind side-by-side human evaluations where voters select the better video generated from matching prompts. Omni Flash shows its strongest results when conditioned on an initial image:

The chart scale begins at 1000 Elo. Omni’s confidence interval spans 1171-1185, making positions 2 through 6 a statistical tie. However, there is a substantial performance gap between Omni and Veo 3.1 (12th place, 1082 Elo), the model Omni officially supersedes.

Hands-On Tests: Coffee Pour, Polish Speech, and a Tram

We tested Gemini Omni Flash 1.1 across multiple standardized prompts via fal.ai in 720p. In our broader benchmark of the best AI video generators (evaluating ten models on September 28), it earned an 8.0/10, tying for 3rd place with Veo 3.1 and FLUX.3 Video. In a street scene prompt featuring a vintage tram (tested October 3), it fell short of Veo 3.1 Lite.

Key takeaway from our testing: Omni’s multilingual speech generation rivals top competitors like H3 Max and Grok Imagine 1.5, but its prompt comprehension can be excessively literal. Camera direction cues risk being turned into physical set props. Instead of typing “tripod”, write “the camera does not move”. Full test logs are available in our AI video generators ranking and our Veo overview.

Gemini Omni Flash vs. Vidu Q4, MiniMax H3, and Seedance

How to choose:

• Best image-to-video fidelity and lowest cost: MiniMax H3 Max takes 1st place and costs $4.80 per minute.
• Complex foreign-language dialogue: Omni Flash or H3 Max - both delivered flawless pronunciation in our speech stress test.
• Extended single-shot clips in 4K: Vidu Q4 Preview handles up to 16-second single-pass shots, though it lacks a pure text-to-video mode.
• Pure text-to-video generation: Omni ranks only 8th; MiniMax H3 or Wan 3.0 produce stronger results from text prompts alone.
• Simple subscription over API billing: Omni Flash is built directly into consumer Gemini subscriptions.

For side-by-side head-to-heads among alternative models, read our comparison of Vidu Q4 vs. Seedance 2.0 vs. MiniMax H3, or inspect identical test prompts in the Video Arena.

Who Is It For?

• Google AI subscribers: Omni is embedded into the Gemini web and mobile interface, requiring no API configuration.
• Creators needing clean voice and dialogue: In our testing, complex spoken phrases generated cleanly without slurring or hallucinated syllables.
• Developers migrating from Veo 3.1 preview: Omni is the designated production replacement, matching Veo 3.1 Fast pricing at 720p.
• Not ideal for European users modifying personal footage: Editing user-uploaded videos remains disabled across the EEA and the UK.

Frequently Asked Questions

Is Gemini Omni Flash free to use?

Not inside the Gemini app: generating video requires a Google AI Plus, Pro, or Ultra subscription. In Google Flow, free accounts receive 50 daily credits, and an Omni 360p clip costs 4-5 credits.

Where is Gemini Omni Flash available?

It is globally available in the Gemini app for paid subscribers, in Google Flow, and through the Gemini API. However, extending or editing user-uploaded video files is blocked in the EEA, Switzerland, and the UK due to local regulatory compliance.

How does Gemini Omni Flash differ from Veo?

Omni supports longer scene extensions (up to 40 seconds), cheap 360p draft renders, conversational chat-based editing, and scores higher in image-to-video benchmarks (1178 vs. 1082 Elo for Veo 3.1). Veo retains its dedicated Quality mode and continues running in Flow and Vertex AI. Explore our full comparison in the Veo hub.

Can Gemini Omni speak languages other than English?

Yes. In our stress test using a notoriously difficult Polish tongue twister, Omni synthesized accurate, natural pronunciation without phonetic errors. Note that Google officially recommends English for prompt descriptions, but dialogue enclosed in quotes handles other languages well.

How much does a minute of Gemini Omni video cost?

Through the API, 720p video with audio costs roughly $6.00 per minute ($0.10/s). A 360p draft runs about $1.80 per minute, while 4K output costs approximately $18.00 per minute.

Is there a Gemini Omni Pro model?

Not yet. Google announced Omni Pro at I/O 2026, confirming it will launch once it demonstrates a substantial performance leap over Flash.

Is Gemini Omni available on OpenRouter?

No, as of October 9, 2026, Omni is not hosted on OpenRouter. Third-party cloud access is available through fal.ai and Replicate.

// read moreGemini Omni 1.2 Flash Review: 15-Second Continuous Shot TestBenchmarksGemini Omni 1.2 vs 1.1 Flash: Head-to-Head Video TestBenchmarksGemini Omni 1.2 Camera Movement: FPV Drones and 360 SpinsBenchmarks