Google video model dates: Omni GA, Veo shutdowns, one table

As of October 3, 2026, Google's deprecations page lists gemini-omni-1.1-flash as GA since August 27 with no shutdown announced, and lists all three veo-3.1 preview ids (standard, fast, lite) for shutdown on October 22, 2026, with gemini-omni-1.1-flash as the replacement. This page collects those dates in one table so a deploy checklist can use it.

Which Google video models have a date?

Everything below comes from Google's own pages, read today. The Google Cloud post adds that on its Agent Platform, Gemini Omni Flash entered Public Preview on June 30, 2026, which is a different surface and a different clock from the Gemini API.

The deprecations page also lists veo-3.0-generate-001, veo-3.0-fast-generate-001 and veo-2.0-generate-001 as shut down June 30, 2026, and two veo-3.0 preview ids as shut down November 12, 2025.

What changes in a request when Veo goes?

Veo 3.1 prices on the pricing page are per second with audio: Standard $0.40 at 720p and 1080p and $0.60 at 4K, Fast $0.10, $0.12 and $0.30, and Lite $0.05 and $0.08 with no 4K. Omni 1.1 Flash is billed on tokens, with video output at $17.50 per million, which the page converts at 5,792 tokens per second of 720p video.

Veo tiers also offered portrait and landscape. Omni lists 16:9 and 9:16, 3 to 10 second generations, and 360p, 720p, 1080p and 4K, where 1080p and 4K are upscaled. A clip longer than 10 seconds is no longer a single request.

How does this map to Sume?

Sume's catalog lists gemini-omni-flash-1.1, not the Veo ids, so there is nothing to migrate on the Sume side. The Sume docs describe the gemini-omni-flash-1.1 row as 3 to 10 seconds at 360p, 720p, 1080p and 4K with native synced audio, and billing of provider list times 1.25 per output second by resolution. Check GET /v1/videos/models for what the catalog reports today, as limits are per model.

If your code calls Google directly, grep for the three preview ids before October 22 and change the model string, then re-test prompts: a Veo prompt tuned to one tier is not guaranteed to behave the same on Omni.

grep -rnE "veo-3\.1-(generate|fast-generate|lite-generate)-preview" src/ config/
# replace each hit with: gemini-omni-1.1-flash (Gemini API)
# or: gemini-omni-flash-1.1 (Sume catalog id)

What should I re-test?

Re-test duration (10 seconds is the Omni ceiling per generation), aspect ratio, resolution, and anything that relied on a Veo-only option. The Veo-to-Omni notes cover the field mapping. Dates can move, so re-read the deprecations page before you freeze a release.