本文介绍了 Google 最新的高保真视频生成模型 Veo 3，该模型现已通过 Gemini API 和 Vertex AI 提供付费预览。Veo 3 是首个集成高保真视频输出和原生音频的 Google 视频模型，支持文本生成视频，未来还将支持图像生成视频。它拥有同步声音、电影级质量和逼真物理模拟等功能。本文展示了早期开发者的采用情况，并提供了 Cartwheel（使用 Veo 3 进行 3D 动画）和 Volley（用于游戏内过场动画）的示例。它提供了快速入门指南，包括定价（每秒 0.75 美元）、Python 代码示例以及指向 Google AI Studio 中的文档和入门应用程序的链接。Google AI 订阅者可通过 Gemini 应用和 Flow 访问 Veo 3，企业客户则可通过 Vertex AI 访问。此外，它还强调了负责任的 AI 开发，并指出所有生成的视频都包含数字 SynthID 水印。
Starting today, we're bringing Veo 3 to developers in paid preview via the Gemini API and Vertex AI.
First unveiled at Google I/O 2025, people around the world have already generated tens of millions of high-quality videos with Veo 3 (along with some new fun and interesting video trends). It is our first video model to incorporate high-fidelity video outputs and native audio, first with text-to-video and soon with image-to-video.
Link to Youtube Video (visible only when JS is disabled)
Developers are already experimenting with Veo 3, discovering how the model can help them brainstorm content, rapidly iterate, and be more efficient.
- Cartwheel developed a system that can take 2D videos of humans and translate it into fully production ready 3D animation on rigged characters. Cartwheel uses Veo 3 to generate realistic, fluid human actions that Cartwheel can then turn into 3D animations for customers.
- Volley uses Veo 3 to produce in-game video cut-scenes that advance the story. With Veo 3, Volley designers can rapidly iterate on the game to deliver the best possible output for an upcoming RPG game called Wit's End.
Veo 3 capabilities
Veo 3 is designed to handle a range of video generation tasks, from cinematic narratives to dynamic character animations. With Veo 3, you can create more immersive experiences by not only generating stunning visuals, but also audio like dialogue and sound effects.
- Synchronized Sound: Natively generates rich audio—dialogue, effects, and music—and synchronizes it with video in a single pass.
- Cinematic Quality: Produces stunning, high-definition video that captures creative nuances in your prompt, from intricate textures to subtle lighting effects.
- Realistic Physics: Simulates real-world physics for authentic motion, from natural character movement to the accurate flow of water and casting of shadows.
Let’s take a look at some examples.
Link to Youtube Video (visible only when JS is disabled)
Link to Youtube Video (visible only when JS is disabled)
Explore these examples and more with Veo 3 in Google AI Studio, available as an SDK template and interactive Starter App to remix, copy and extend. The Starter App and its sample code offer a convenient way for Paid Tier users to rapidly prototype with Veo 3 and more on the Gemini API, directly from Google AI Studio.
Click the Key button in the top right of the AI Studio Build interface to select a Google Cloud Project with billing enabled to use the Paid Tier in AI Studio apps. See the FAQs for more.
Get started with Veo 3 in the Gemini API
Veo 3 will be priced at $0.75 per second for video and audio output. Additionally, Veo 3 Fast will be available soon, offering a faster and more cost-effective option for video creation.
Here’s a basic Python example to create a video:
import time
from google import genai
from google.genai import types
client = genai.Client()
operation = client.models.generate_videos(
model="veo-3.0-generate-preview",
prompt="a close-up shot of a golden retriever playing in a field of sunflowers",
config=types.GenerateVideosConfig(
negative_prompt="barking, woofing",
),
)
# Waiting for the video(s) to be generated
while not operation.done:
time.sleep(20)
operation = client.operations.get(operation)
generated_video = operation.result.generated_videos[0]
client.files.download(file=generated_video.video)
generated_video.video.save("veo3_video.mp4")Building responsibly with Veo 3 in the Gemini API
All videos generated by Veo 3 models will continue to include a digital SynthID watermark. To get started, check out the documentation, cookbook, and a Veo 3 starter app in Google AI Studio:
- Try the Veo 3 starter app (paid tier only)
In addition to being available via the Gemini API in Google AI Studio, Veo 3 is also available to Google AI subscribers in the Gemini app and Flow, and to enterprise customers via Vertex AI.
