news

Google turns off the three Veo 3.1 preview endpoints today, with Gemini Omni 1.1 Flash as the only named replacement

The Gemini API documentation dates the shutdown of veo-3.1-generate-preview, its Fast and Lite siblings and gemini-omni-flash-preview to October 22, 2026, four months after the Veo 3.0 and 2.0 IDs went dark.

By Daniel Ochoa · 22 October 2026

Today is the date Google published for the end of Veo 3.1 on the Gemini API. The video generation documentation carries a warning that veo-3.1-generate-preview, veo-3.1-fast-generate-preview and veo-3.1-lite-generate-preview are deprecated and will be shut down on October 22, 2026, and it points developers to Gemini Omni Flash (gemini-omni-1.1-flash) or to the generally available models on the Gemini Enterprise Agent Platform to avoid interruption.

The same date appears on the per-model page for veo-3.1-generate-preview, so anyone who reached the model through Google's own developer docs rather than a third-party host has had the notice in front of them for months.

Four preview IDs retire together, all pointing at one model

The Veo previews are not going alone. Google's deprecations page lists gemini-omni-flash-preview on the same October 22 date, which means the preview build of the model family that replaced Veo retires on the day the Veo previews do. All four IDs name gemini-omni-1.1-flash as the replacement, a model Google's deprecations listing dates to August 27, 2026 and currently carries no shutdown date of its own.

The wording differs depending on which Google page you read. The video docs say the previews will be shut down on October 22. The deprecations page frames published dates as the earliest point at which a model can be turned off, with the exact date announced separately, so a request against one of these IDs answering after today would not contradict the notice.

The replacement is not a model string swap

Moving from Veo 3.1 to Omni Flash changes the shape of the call, not just the identifier. Google's video documentation describes Omni Flash as turning prompts and images into short videos and refining results across multiple turns through the Interactions API, which is a different route from the long-running video operation that the Veo preview IDs served. Anything built around the Veo endpoint has to be rewritten rather than repointed.

In our price radar, this is what the Veo 3.1 route costs across the platforms that still carry it:

And the same view for the model Google names as the landing spot, Gemini Omni:

The second forced Veo migration of 2026

Developers who are migrating today already did this in the summer. Google's deprecations listing records veo-2.0-generate-001, veo-3.0-generate-001 and veo-3.0-fast-generate-001 as shut down on June 30, 2026, and the Veo 3.1 previews were the destination at the time. Those destinations are the IDs expiring now, which gives the Gemini API Veo lineage a working life of roughly four months per generation.

What survives is the enterprise route. Google's migration note names the generally available models on the Gemini Enterprise Agent Platform, the renamed Vertex AI, as an alternative to Omni Flash, and the Vertex-side Veo identifiers follow a separate lifecycle with no announced end date. The split matters for anyone comparing quotes: the same model name can be live on one Google surface and withdrawn on another on the same morning.

It is also the second major video API closure of the autumn, after OpenAI's Sora API date passed on September 24. Two of the three best-known Western video endpoints have now closed their developer-facing preview routes within a month of each other.

Sources: Google AI for Developers, Gemini API deprecations, Veo 3.1 model page.

Daniel Ochoa: Covers model launches, shutdowns and pricing changes as they happen. Reads deprecation notices for a living so you do not have to.