Google Veo 3
Google's Veo 3. High-fidelity text-to-video with native audio generation, up to 8s clips.
Google Veo 3 is video generation AI model from Google DeepMind, priced at €0.000 per 1M input tokens with a unknown context window.
Examples
See what Google Veo 3 can generate
Cinematic
"A dramatic slow-motion shot of a bird diving toward water"
Pricing
API Integration
Use our OpenAI-compatible API to integrate Google Veo 3 into your application.
npm install railwailimport railwail from "railwail";
const rw = railwail("YOUR_API_KEY");
// Simple — just pass a string
const reply = await rw.run("veo-3", "Hello! What can you do?");
console.log(reply);
// With message history
const reply2 = await rw.run("veo-3", [
{ role: "system", content: "You are a helpful assistant." },
{ role: "user", content: "Explain quantum computing simply." },
]);
console.log(reply2);
// Full response with usage info
const res = await rw.chat("veo-3", [
{ role: "user", content: "Hello!" },
], { temperature: 0.7, max_tokens: 500 });
console.log(res.choices[0].message.content);
console.log(res.usage);Deep dive — Google DeepMind's Google Veo 3
Google DeepMind is the merged AI research organisation formed in April 2023 from Google Brain and DeepMind under Demis Hassabis. DeepMind itself was founded in 2010 by Hassabis, Shane Legg and Mustafa Suleyman and acquired by Google in 2014. The Veo programme delivered Veo 1 (May 2024, Google I/O), Veo 2 (December 2024) and Veo 3 (May 2025, Google I/O). Veo 3 was the headline announcement of I/O 2025 and was widely described as the first major commercial video model to ship with natively generated synchronized audio -- music, ambient sound effects and spoken dialogue with lip-sync -- in addition to high-fidelity 1080p video. Veo 3 is available via Vertex AI, the Gemini API, Google Labs (VideoFX, Whisk) and the Flow filmmaking surface for creators.
Visit Google DeepMind →Veo 3 extends DeepMind's Veo 2 video diffusion architecture with a joint audio-video diffusion stack. Video is encoded into a learned spatio-temporal latent space and denoised by a transformer-based diffusion model conditioned on Gemini-family text embeddings and optional image embeddings. A coupled audio diffusion module produces a synchronized soundtrack consisting of music, ambient sound, sound effects and short spoken dialogue with lip-sync. Veo 3 generates clips up to 8 seconds natively at 1080p (and 4K via cascaded super-resolution), with extensions for longer sequences. Cinematographic prompt language is richly understood (lens, lighting, rig, shot scale). Veo 3 is the first Veo to compose synchronized audio at this fidelity, marking the major industry milestone of 'audio-on-by-default' video generation. Training uses a curated multilingual audio-video corpus including licensed footage, public web data and YouTube under Google's terms.
- Parameters
- Undisclosed
- Context
- unknown
- First major commercial video model with native synchronized audio (music, SFX, dialogue with lip-sync)
- Up to 8-second 1080p clips natively, up to 4K via cascaded super-resolution
- Text-to-video and image-to-video
- Rich cinematographic prompt vocabulary (lenses, lighting, camera rigs)
- Strong physical plausibility and object permanence
- Multilingual prompts via Gemini text encoders
- Available via Vertex AI, Gemini API, VideoFX, Whisk and Flow
- SynthID audio + visual watermarking embedded in every output
- Best for: audio-on-by-default cinematic shorts, ads with sound, music-video creative.
Curated multilingual audio-video corpus including licensed footage, public web video (including YouTube under Google's terms) and synthetic data, with multi-granularity captions produced by Gemini vision-language models. Exact size undisclosed.
License: Proprietary commercial licence via Google Cloud / Vertex AI and the Gemini API; commercial use permitted under Google's generative-AI terms; SynthID watermark mandatory on all outputs.
Known limitations
- 8-second native clip limit
- Audio is short-form and English-leaning
- Strict moderation on people, brands and political content
- Closed access with gated waitlist on some surfaces
- Closed model with no full technical paper
Frequently asked questions
Related Models
View all Video GenerationGoogle Veo 2
Google's state-of-the-art video generation model. Simulates real-world physics with various visual styles.
Google Veo 3.1
Latest Veo with image-to-video and context-aware audio
Kling v3
Cinematic video up to 15s with multi-shot and native audio
Kling v3 Omni
Most versatile: multi-reference images, video editing, native audio
Start using Google Veo 3 today
Get started with free credits. No credit card required. Access Google Veo 3 and 100+ other models through a single API.