Google Veo 3 Video API - Runcrate

Veo 3 generates video with synchronized audio in a single pass. Ambient sounds, dialogue, and effects are inferred from your prompt — no separate audio pipeline.

Available models

Model	Audio	Speed	Best for
Veo 3.0 Audio	Auto-generated	Medium	Videos with sound
Veo 3.0 Fast Audio	Auto-generated	Fast	Quick drafts with sound
Veo 2.0	None	Medium	Silent video, lower cost

Video with audio

from runcrate import Runcrate

client = Runcrate(api_key="rc_live_YOUR_API_KEY")
job = client.models.generate_video_and_save(
    "cafe-scene.mp4",
    model="google/veo-3.0-audio",
    prompt="Barista pouring latte art, espresso machine hissing, cups clinking, ambient cafe chatter",
    duration=6,
    on_status=lambda j: print(f"  {j.status}..."),
)

import Runcrate from '@runcrate/sdk';

const rc = new Runcrate({ apiKey: 'rc_live_YOUR_API_KEY' });
const job = await rc.models.generateVideo({
  model: 'google/veo-3.0-audio',
  prompt: 'Barista pouring latte art, espresso machine hissing, cups clinking',
  duration: 6,
});
let status = job;
while (status.status !== 'completed' && status.status !== 'failed') {
  await new Promise(r => setTimeout(r, 5000));
  status = await rc.models.getVideoStatus(job.id);
}
await Bun.write('cafe-scene.mp4', await rc.models.downloadVideo(job.id));

curl https://api.runcrate.ai/v1/videos \
  -H "Authorization: Bearer rc_live_YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "google/veo-3.0-audio", "prompt": "Barista pouring latte art, espresso machine hissing", "duration": 6}'

All examples below use the same client from above.

Fast audio drafts

client.models.generate_video_and_save(
    "draft.mp4", model="google/veo-3.0-fast-audio",
    prompt="Dog running through autumn leaves, leaves crunching, birds chirping",
    duration=4, on_status=lambda j: print(f"  {j.status}..."),
)

Silent video (Veo 2)

When adding your own audio in post-production:

client.models.generate_video_and_save(
    "mountains.mp4", model="google/veo-2.0",
    prompt="Aerial drone shot through misty mountain peaks at sunrise, cinematic color grading",
    duration=8, on_status=lambda j: print(f"  {j.status}..."),
)

Audio prompting

Be explicit about sounds in your prompt:

client.models.generate_video_and_save(
    "storm.mp4", model="google/veo-3.0-audio",
    prompt="Lightning over a dark ocean, deep rumbling thunder, rain hitting water, wind howling",
    duration=8, on_status=lambda j: print(f"  {j.status}..."),
)

Words like “hissing”, “clinking”, “rustling”, and “crackling” in your prompt help the model generate more accurate audio.

Tips

Audio is automatic. The audio variants produce video with matching sound in one pass.
Describe sounds explicitly. “Rain hitting a window” beats “rainy scene” for audio quality.
Veo 2 for silent. If adding your own music/voiceover, use Veo 2.0 to avoid paying for unused audio.
Duration. All Veo models support 4, 6, and 8 second durations.