Google Veo

innFactory AI Consulting from Rosenheim advises companies in the DACH region on GDPR-compliant integration of video AI. Google Veo 3.1 is Google’s video model for text-to-video and image-to-video with native audio; per Google’s documentation it runs on the Gemini Enterprise Agent Platform (formerly Vertex AI) only in the US region us-central1, and Google does not offer an EU region or EU data residency for Veo. Google also offers Gemini Omni 1.1 Flash, a conversational model for video generation and editing. Veo 2 and Veo 3 are shut down, and no Veo 4 has been announced. As of October 2026.

Current models at a glance

Veo 3.1 (Standard, Fast, Lite)

Veo 3.1 generates videos from text or images in 720p, 1080p and 4K, with native audio generation (synchronised dialogue, sound effects, music). It supports text-to-video, image-to-video, videos from a first and last frame, reference images (“Ingredients to Video”, up to three images for character and style consistency) and extension of existing videos. A generation lasts 4, 6 or 8 seconds; reference-image-to-video supports only 8 seconds per the model page. Besides 16:9, native portrait 9:16 is possible.

Veo 3.1 Fast is the faster and cheaper variant, Veo 3.1 Lite (veo-3.1-lite-generate-001, preview since 2 April 2026) the fastest and cheapest. Per the model pages, all three models run only in us-central1. All videos generated by Veo carry a SynthID watermark, an invisible signature that makes AI-generated content verifiable and helps with transparency obligations. Per the model page, Veo 3.1, Fast and Lite also support Content Credentials (C2PA).

Gemini Omni 1.1 Flash

Gemini Omni 1.1 Flash (gemini-omni-1.1-flash, GA in the Gemini API since 27 August 2026) is not a classic text-to-video model but works conversationally: after a first generation, scenes can be adjusted without re-prompting. Per Google, the model supports video extension (continuing a clip), interpolation between two images (first and last frame) and a resolution parameter with 360p, 720p (default), 1080p and 4K; 1080p and 4K are produced by upscaling. Input videos for editing and extension may be at most 10 seconds long, and extensions generate a continuation of 3 to 10 seconds. Every generation receives a SynthID watermark. On the Agent Platform, the variant gemini-omni-1.1-flash-preview (preview, region global) is listed. For its predecessor gemini-omni-flash-preview, the Agent Platform model page states a deprecation date of 6 November 2026.

Pricing

Veo prices come from the pricing page of the Gemini Enterprise Agent Platform. For Omni, the pricing pages of the Gemini API and the Agent Platform list the same token prices; per the Agent Platform pricing page, one second of video output equals 1,931 tokens (360p), 5,792 tokens (720p), 8,688 tokens (1080p) or 17,376 tokens (4K). As far as we know, Google offers no formal SLAs for Veo.

EU availability and GDPR

Neither an EU single region (including europe-west3/Frankfurt) nor an EU multi-region or global endpoint is documented for Veo. For GDPR-relevant video production this means that input images and prompts are processed in the USA; the transfer must be secured contractually (for example standard contractual clauses) and through a data protection impact assessment. Note also that certain functions, in particular generating depictions of people, may be restricted in the EU and UK. As long as no personal or confidential content ends up in prompts and reference images, use for marketing and concept visualisation is practical.

Migration and breaking changes

• Gemini API previews: veo-3.1-generate-preview, veo-3.1-fast-generate-preview, veo-3.1-lite-generate-preview and gemini-omni-flash-preview have been shut down since 22 October 2026. Google names gemini-omni-1.1-flash as the replacement. Those who want to stay with Veo characteristics switch to the GA models on the Agent Platform (veo-3.1-generate-001, veo-3.1-fast-generate-001), which run only in us-central1.
• Veo 2 and Veo 3: veo-2.0-generate-001, veo-3.0-generate-001 and veo-3.0-fast-generate-001 have been shut down since 30 June 2026. The Agent Platform names Veo 3.1 as the upgrade; the Gemini API names gemini-omni-1.1-flash or the GA models on the Agent Platform.
• API change to Omni: gemini-omni-1.1-flash uses its own task parameters in the Gemini API (such as extend, image_to_video) and a resolution parameter; existing Veo calls cannot be carried over unchanged.

Model lifecycle

“No earlier than” follows Google’s note that Agent Platform retirement dates are not brought forward but may be extended. Google has not announced a Veo 4.

Integration with CompanyGPT

For companies building multimodal AI workflows, Veo can be combined with CompanyGPT to steer text, image and video generation in a GDPR-compliant setup. Because Veo processes exclusively in the USA, we recommend clear rules on which content may be used as a prompt or reference image. The innFactory AI Gateway lets you control budgets and access for video models centrally.

Our recommendation

• Production use with Veo characteristics: Veo 3.1 (Standard or Fast) on the Agent Platform, for marketing, product demos and creative prototypes. Keep an eye on the retirement date (no earlier than 17 November 2026) and Google’s replacement.
• Conversational editing and extension: Gemini Omni 1.1 Flash via the Gemini API; the Agent Platform variant is still in preview.
• Inexpensive drafts: Veo 3.1 Lite (preview) for storyboards and iterations.
• Strict EU data residency or on-premise: Veo is not suitable; evaluate open-source alternatives, even if their quality does not need to match Veo.
• Before you start: secure the US processing contractually, keep personal content out of prompts and reference images, and plan for SynthID labelling and EU AI Act transparency duties for synthetic media.