SunsettingAI Video

Gemini Omni Flash Preview (gemini-omni-flash-preview)

by Google

Announced

2026-08-27

Complete

2026-10-22

Refund window closes

—

Refund status

No refund applicable

Google released `gemini-omni-flash-preview`, its conversational video generation and editing model, on June 30, 2026. On August 27, 2026 the Gemini API changelog announced the generally available version, `gemini-omni-1.1-flash`, and said the preview endpoint would be deprecated on September 30, 2026. The Gemini API deprecations page (last updated October 9, 2026) now lists October 22, 2026 as the preview's shutdown date, with `gemini-omni-1.1-flash` as the replacement. Google notes that listed shutdown dates are the earliest possible dates and that it will confirm the exact date in advance. October 22 is also the date the three Veo 3.1 preview models are listed to shut down, with the same replacement, so any Gemini API video pipeline still on a preview model string has the same deadline. The pricing page lists identical prices for the preview and the GA model, so this move is mostly a model string change that also unlocks new features.

Refund flow

1
Not applicable. gemini-omni-flash-preview is a pay-per-token API model on the paid tier, with no pre-paid balance tied to the model string, so there is no refund category to open.
2
If you are billed for gemini-omni-flash-preview after it is shut down, treat it as a billing error and raise it through Google Cloud billing support with the invoice line, rather than as a refund request.
3
No re-budget is needed for the switch itself. Google's Gemini API pricing page on October 12, 2026 lists the same paid tier prices for gemini-omni-flash-preview and gemini-omni-1.1-flash: $1.50 per 1M input tokens (text, image, video or audio), $9.00 per 1M text output tokens and $17.50 per 1M video output tokens. Neither model is available on the free tier.
4
In clip terms: Google bills video at 5,792 output tokens per second of 720p video and says that works out to about $0.10 per second. At $17.50 per 1M tokens that is $0.101 per second, so a 3 second clip is about $0.30 and a 10 second clip, the maximum length, is about $1.01, before input tokens. The pricing note is stated for 720p only. The page does not list a separate rate for the upscaled 1080p and 4K outputs that the GA model adds, so check your first invoice before running those at volume.

Migration path

Recommended successor

Gemini Omni Flash (gemini-omni-1.1-flash)

Why this is the closest fit

It is the GA version of the same model. Google's Gemini Omni Flash model page lists gemini-omni-1.1-flash as the stable version and gemini-omni-flash-preview as the preview version of one model, and the deprecations page names it as the only replacement. The Omni guide shows text to video, image to video, conversational edits chained with previous_interaction_id through the Interactions API, and editing of uploaded videos, all on gemini-omni-1.1-flash.

What differs from the original

The August 27, 2026 changelog lists three capabilities as new in the GA release. First, video extension: the model can generate a 3 to 10 second continuation at the end of a clip, either one it generated earlier in the conversation or an uploaded video of 10 seconds or less. It only appends to the end, it cannot extend an uploaded clip where someone is talking to add more dialogue, and extending uploaded videos is not available in the EEA, Switzerland or the UK. Second, interpolation: image to video accepts up to 2 images, a first and a last frame, and generates the transition between them. Third, resolution control: output can be 360p, 720p (the default), 1080p or 4K, and Google says the 1080p and 4K outputs are upscaled rather than generated natively. Limits that stay the same per the model page and guide: clips are 3 to 10 seconds at 24 FPS, landscape 16:9 is the default with 9:16 available for portrait, every video carries an invisible SynthID watermark, negative prompts are not a separate field, and YouTube videos cannot be used as a source.

Alternatives by use case

You call gemini-omni-flash-preview for short text to video or image to video clips

Change the model string to gemini-omni-1.1-flash before October 22, 2026 and re-run a few of your real prompts. The listed price is the same, so the only thing to check is the output itself

You stitch several Omni clips together to get past the 10 second limit

Try the GA extension feature, which continues a clip from its last frame, instead of joining separate generations, then compare the seams against your current cut

You need a clip to land on a specific end frame, such as a product shot or logo card

Use the GA first and last frame interpolation with up to 2 images, which the preview changelog entry did not include

You are also on a Veo 3.1 preview model in the Gemini API

Both shut down on October 22, 2026 with gemini-omni-1.1-flash as the replacement, so plan one migration for both rather than two

What this tool meant

gemini-omni-flash-preview will have lasted under four months, from a June 30, 2026 preview launch to an October 22, 2026 shutdown, with the GA model arriving after two of them. That is a short window, and it is the normal cost of building on a preview string: the model was always going to be replaced by a stable one. The useful detail is that Google kept the price flat between preview and GA, which is not what happened with its TTS previews, where the GA replacements get a scheduled price increase on January 1, 2027. For an AI video workflow, the October 22 date matters more than the model itself, because it is the day both Omni preview and Veo 3.1 preview strings stop working in the Gemini API.

Sources

Spent credits on AI video that failed?

AVA automates the refund flow across every video provider

Free Chrome extension. Detects the failure mode, captures evidence, drafts the refund email with the technical term and Generation ID. Click send.

Install Free Extension →Browse 94 Failure Modes