Veo 2.0 and Veo 3.0 shut down June 30: what model id to call now

If your Veo call returns an error since June 30, the model id was retired. Google's changelog lists veo-2.0-generate-001, veo-3.0-generate-001 and veo-3.0-fast-generate-001 as shut down on 2026-06-30 (read 2026-10-03). The current Veo page lists three ids: veo-3.1-generate-preview, veo-3.1-fast-generate-preview and veo-3.1-lite-generate-preview.

Google's video overview now recommends Gemini Omni Flash as the default and keeps Veo 3.1 for native-audio and extension-specific work (read 2026-10-03). Sume does not list a Veo model, so the Sume route is a model swap, not a rename.

Retired ids and what replaces them

The pairing in the middle column is the nearest tier by name, not a Google-stated mapping. Sume ids come from the model list your key returns.

Moving the call to Sume

Sume's POST /v1/videos takes a bare model id, a prompt, and optional duration, resolution and aspect_ratio. Ask GET /v1/videos/models which durations and resolutions each id accepts before you submit, because limits differ by model.

Veo's 4, 6 and 8 second durations do not map onto every Sume model. Omni Flash on Sume takes 3 to 10 seconds; seedance-2.5 takes 4 to 30. Check the field-level differences in the parameter map post.

What to do this week

A retired id fails the same way every time, so this is a one-time fix rather than a retry problem.

• Grep your code for the three retired ids.
• Choose a replacement from the capability list, not from a blog headline.
• Run one clip through the new model and compare duration, resolution and audio before cutting traffic.