Build with Veo 3, Now Available in the Gemini API

As of today, Veo 3 is rolling out to developers in paid preview through the Gemini API and Vertex AI.

Since its debut at Google I/O 2025, Veo 3 has already powered tens of millions of high-quality videos generated by people around the world, along with a wave of fresh video trends. It is Google's first video model to combine high-fidelity output with native audio, starting with text-to-video and adding image-to-video soon.

Developers are already putting Veo 3 to work to brainstorm content, iterate quickly, and move faster:

• Cartwheel built a pipeline that converts 2D videos of people into production-ready 3D animation on rigged characters. It leans on Veo 3 to generate realistic, fluid human motion, which Cartwheel then translates into 3D animations for its customers.
• Volley uses Veo 3 to produce in-game cut-scenes that move the story forward. The team rapidly iterates on these sequences to land the best possible result for an upcoming RPG, Wit's End.

What Veo 3 can do

Veo 3 is built to handle a wide span of video generation work, from cinematic narrative to dynamic character animation. It generates not only the visuals but also matching audio such as dialogue and sound effects.

• Synchronized sound: Generates rich audio natively, including dialogue, effects, and music, and syncs it to the video in a single pass.
• Cinematic quality: Delivers high-definition video that captures the nuance in your prompt, from fine textures to subtle lighting.
• Realistic physics: Models real-world physics for believable motion, from natural body movement to accurate water flow and shadow casting.

A couple of examples follow.

You can explore these and more inside Google AI Studio, available as an SDK template and an interactive Starter App you can remix, copy, and extend. For Paid Tier users, the Starter App and its sample code are a quick way to prototype with Veo 3 and other models on the Gemini API directly from AI Studio.

To enable the Paid Tier in AI Studio apps, click the Key button in the top right of the Build interface and select a Google Cloud project with billing turned on.

Getting started in the Gemini API

Veo 3 is priced at $0.75 per second of video and audio output. A faster, more cost-effective option, Veo 3 Fast, is also coming soon.

Here is a minimal Python example that creates 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",
    ),
)

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

Every video produced by Veo 3 models continues to carry a digital SynthID watermark. To get going, consult the documentation, the Veo cookbook, and the Veo 3 starter app in Google AI Studio (paid tier only).

Beyond the Gemini API in AI Studio, Veo 3 is also available to Google AI subscribers in the Gemini app and in Flow, and to enterprise customers through Vertex AI.

Example prompts

Prompt 1 — Fluffy Characters Stop Motion: Inside a brightly colored, cozy kitchen made of felt and yarn. Professor Nibbles, a plump, fluffy hamster with oversized glasses, nervously stirs a bubbling pot on a miniature stove, muttering, "Just a little more... 'essence of savory,' as the recipe calls for." The camera holds a mid-shot, capturing his frantic stirring. Suddenly the pot lets out a loud "POP!" followed by a comical "whoosh," and a geyser of iridescent green slime erupts, covering the entire kitchen. Professor Nibbles shrieks, "Oh, dear! Not again!" and scurries off, leaving a trail of tiny, panicked squeaks.

Prompt 2: The sequence opens on an extreme close-up of a single gear, slowly turning and catching harsh sunlight. The camera pulls back in one continuous move, revealing that this is just one part of a colossal mechanical heart half-buried in a desolate, rust-colored desert. A sweeping aerial shot establishes its enormous scale and isolation. The camera descends to catch pipes hissing steam and the rhythmic thump that echoes across the empty plains, a subtle shake synced to each beat. A lateral tracking shot finds tiny robed figures scurrying over the metallic surface, and the camera follows one as it polishes brass valves and tightens immense bolts. A complex movement circles the whole structure, capturing maintenance teams working in precarious positions across the rusted exterior. The final shot starts tight on one tiny figure's meticulous work before a dramatic pull-out reveals the true scale of the heart and the minuscule caretakers tending the vital organ of an unseen, sleeping giant that extends beyond the frame.