SunsettingAI Video

Veo 3.1 Preview (veo-3.1-generate-preview, veo-3.1-fast-generate-preview, veo-3.1-lite-generate-preview)

by Google

Announced

—

Complete

2026-10-22

Refund window closes

—

Refund status

No refund applicable

Google's Gemini API deprecations page (last updated October 1, 2026) lists all three Veo 3.1 preview model IDs, `veo-3.1-generate-preview`, `veo-3.1-fast-generate-preview` and `veo-3.1-lite-generate-preview`, with a shutdown date of October 22, 2026 and `gemini-omni-1.1-flash` as the recommended replacement. Google notes that listed shutdown dates are the earliest possible dates. This is the second Veo migration in four months: the GA models `veo-2.0-generate-001`, `veo-3.0-generate-001` and `veo-3.0-fast-generate-001` were shut down on June 30, 2026, and the release notes told those users to move to the Veo 3.1 preview IDs that are now being retired. The replacement is not a drop-in model string. Gemini Omni Flash is called through the Interactions API, produces 3 to 10 second clips instead of Veo's 4, 6 or 8, and generates 1080p and 4K by upscaling. Whether the switch costs more or less depends on which Veo tier you used.

Refund flow

1
Not applicable. Veo in the Gemini API is billed per second of generated video on the paid tier, with no pre-paid balance tied to a model ID, so there is nothing to refund when the model is retired.
2
If you see charges for veo-3.1 preview generations dated after the model stops serving, treat it as a billing error and raise it through Google Cloud billing support with the request details. Google's pricing page states you are only charged when a video is successfully generated.
3
Re-budget per tier before you switch, because the prices on Google's pricing page on October 5, 2026 move in different directions. Veo 3.1 Standard is $0.40 per second at 720p and 1080p and $0.60 at 4K. Veo 3.1 Fast is $0.10 at 720p, $0.12 at 1080p and $0.30 at 4K. Veo 3.1 Lite is $0.05 at 720p and $0.08 at 1080p, with no 4K. Gemini Omni Flash is billed by output tokens at $17.50 per 1M video tokens, at 5,792 tokens per second of 720p video, which Google states works out to about $0.10 per second at 720p. So at 720p, Standard users pay about 75% less, Fast users pay about the same, and Lite users pay about twice as much. The pricing page does not state a per-second figure for Omni at 1080p or 4K.

Migration path

Recommended successor

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

Why this is the closest fit

It is the replacement Google names on the deprecation rows for all three Veo 3.1 preview IDs, and it went GA in the Gemini API on August 27, 2026. It covers the same core jobs: text to video, image to video, first and last frame interpolation, video extension, 16:9 and 9:16 output, and video with audio from a text prompt.

What differs from the original

Four things change besides the model name. First, the call: the Omni guide uses client.interactions.create with an input string and returns the clip as base64 in output_video, and edits chain with previous_interaction_id, so code built around the Veo generate-videos request and its parameters needs rewriting, not a string swap. Second, length: Omni outputs 3 to 10 seconds per clip, while Veo 3.1 takes durationSeconds of 4, 6 or 8 and requires 8 for 1080p, 4K, extension and reference images. Third, resolution: Omni supports 360p, 720p (default), 1080p and 4K, and the August 27, 2026 release note says 1080p and 4K are generated using upscaling. Fourth, audio: the Omni guide says uploading audio references is unsupported, any audio in a video reference is ignored, and voice editing is not supported.

Alternatives by use case

You render 720p drafts on Veo 3.1 Standard

Gemini Omni Flash lists about $0.10 per second at 720p against $0.40 on Veo 3.1 Standard. Re-run a few of your real prompts and compare the clips before you move a whole pipeline

You run high-volume clips on Veo 3.1 Lite to keep cost down

Budget for roughly double the per-second cost at 720p on Omni, about $0.10 against $0.05, and check whether 360p output is good enough for drafts

You deliver final masters at 1080p or 4K

Omni produces those sizes by upscaling. Compare a native Veo 3.1 render you already have against an Omni render of the same prompt before October 22, 2026, and look at fine text and edges

Your edit timeline is built on 8 second shots

Omni clips run 3 to 10 seconds, so you can keep 8 second shots, but update any code that sends durationSeconds as "4", "6" or "8"

What this tool meant

Veo 3.1 preview was the escape route Google gave Veo 2 and Veo 3 users in June 2026, and four months later it is on the shutdown list too. In the Gemini API, preview IDs carry a shutdown date, and a team that moved once to stay online now has to move again, this time to a different API shape. If a video pipeline depends on a hosted model, keep the prompt, the seed if you used one, and a reference render for every shot you care about, so you can check the new model against your own footage instead of a demo reel. The Veo 3.1 GA models on Google's Gemini Enterprise Agent Platform are a separate product and were not checked for this record.

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