Veo vs Gemini Omni

What the October 22 Shutdown Means

Google's deprecation schedule makes the Gemini API decision simple. Veo 2 and the Veo 3.0 GA models closed on 2026-06-30. The three Veo 3.1 preview models (standard, Fast and Lite) close on 2026-10-22, with gemini-omni-1.1-flash listed as the replacement. After that date a developer using a Gemini API key has one Google video model to call. The same date also retires the earlier gemini-omni-flash-preview, so anyone who adopted Omni during its June preview needs to move to the 1.1 model ID too.

The enterprise platform keeps Veo alive. There, veo-3.1-generate-001 and veo-3.1-fast-generate-001 are generally available, with retirement listed as "November 17, 2026 or later", and Veo 3.1 Lite is in preview. Teams that need a GA model with enterprise terms can stay on Veo for now, but the retirement language suggests planning a migration rather than building on it. This is the second time in 2026 that a video API has been pulled out from under developers, after OpenAI removed the Sora API in September.

Generation Versus Conversation

Veo is a classic generation model. Each request produces a clip from a prompt, optionally guided by up to three reference images, a first and last frame to interpolate between, or a previous clip to extend in seven-second steps up to 148 seconds at 720p. Its controls act at generation time.

Omni works like a chat. Through the stateful Interactions API, each turn can refer back to the previous one, so a developer can generate a clip and then ask for changes in plain language without starting over. It accepts text, images, video and audio as input, takes up to three short video references and two images, and can edit uploaded clips of up to ten seconds. Scene extension adds ten seconds at a time up to 40 seconds. For products where users refine a shot, this is a different kind of tool, and Google already uses Omni in Flow and YouTube Shorts Remix.

Quality and Resolution

On Artificial Analysis's arenas as fetched on 2026-10-20, Omni 1.1 Flash ranks well above Veo 3.1: ninth against twentieth for text-to-video with audio, and fourth against fourteenth for image-to-video. Neither leads; Alibaba's Wan 3.0, Utopai X and ByteDance's Seedance 2.5 top the text-to-video board (see Text-to-Video vs Image-to-Video).

Resolution is a real caveat for Omni. It generates at 360p or 720p, and Google documents its 1080p and 4K outputs as upscaled rather than generated natively. Veo 3.1 offers 1080p and 4K for eight-second clips, with Lite limited to 1080p. For delivery at high resolution, test both on your material rather than assuming the higher label wins.

Price

Omni is priced like a language model: $17.50 per million video-output tokens, billed at 5,792 tokens per second of 720p video, which Google rounds to about $0.10 a second. Text output costs $9 and input $1.50 per million tokens, and there is no free or batch tier. Google says a 360p draft is up to 60% faster and costs a third as much as 720p, which makes iteration cheap. Google publishes no per-second figure for 1080p or 4K; third-party estimates scale the token count with resolution.

Veo is priced per second. Standard Veo 3.1 costs $0.40 a second at 720p or 1080p and $0.60 at 4K on the Gemini API; Veo 3.1 Fast costs $0.10 to $0.30 a second depending on resolution; and Veo 3.1 Lite $0.05 to $0.08. At 720p, standard Veo costs about four times as much as Omni, while Veo Fast costs about the same. The enterprise platform has its own price list, which was not checked for this page.

Provenance and Safety

Both models embed Google's SynthID watermark in everything they generate, and both generate native audio, which raises the same impersonation questions when it includes voices. Omni's ability to edit uploaded video makes consent for real people in the source footage more important. Platforms that publish this output increasingly require disclosure (see AI content labeling and C2PA).