> ## Documentation Index
> Fetch the complete documentation index at: https://docs.bytespike.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# veo3.1

> Google's Veo 3.1 video model. Native audio generation alongside the video track — useful when your clip is the final output rather than a layer that gets a separate sound-design pass.

`veo3.1` is Google's Veo 3.1 model. Same two-phase task-based protocol as the other video models, with one differentiator worth knowing about: **native audio generation alongside the video track**. The same `submit → poll` flow produces an MP4 with an audio layer the model invented to match the scene — useful for one-shot deliverables that won't get a separate sound-design pass.

**Pricing:** \$0.40 / second of generated footage — see the [rate card](/pricing). Failures don't bill; per-second pricing applies to generated footage length, and audio doesn't add a separate line item on this tier.

## Protocols

| Protocol              | Path                                                     | Purpose                                                    |
| --------------------- | -------------------------------------------------------- | ---------------------------------------------------------- |
| OpenAI Video — submit | `POST https://llm.bytespike.ai/v1/videos/generations`    | enqueues; returns `task_id`                                |
| OpenAI Video — poll   | `GET https://llm.bytespike.ai/v1/videos/tasks/{task_id}` | returns `status`, `result_url`, and `audio_url` when ready |

## Quickstart

```bash theme={null}
TASK_ID=$(curl -s https://llm.bytespike.ai/v1/videos/generations \
  -H "Authorization: Bearer $BYTESPIKE_API_KEY" \
  -H "content-type: application/json" \
  -d '{
    "model": "veo3.1",
    "prompt": "Rain falling on a quiet street at night, distant car passing",
    "duration_seconds": 5,
    "size": "1280x720",
    "audio": true
  }' | jq -r .task_id)

# Poll GET /v1/videos/tasks/{task_id} until status is "succeeded"
# Response includes both result_url (video) and audio_url (audio track)
```

## Capabilities

| Capability                           | Supported             |
| ------------------------------------ | --------------------- |
| Text-to-video                        | ✅                     |
| Image-to-video (with `source_image`) | ✅                     |
| **Native audio generation**          | ✅ (set `audio: true`) |
| `duration_seconds` 5 / 10            | ✅                     |
| `size` 1280×720 / 1920×1080          | ✅                     |
| Modality                             | video                 |
| Capability bucket                    | `video_generate`      |

## When to use

* **One-shot deliverable** — clip is the final output, no sound-design pass coming.
* **Ambient / atmospheric footage** — rain, wind, city noise, where Veo's native audio is more authentic than dubbing-over-silent footage.

When **not** to use:

* You already have your own sound design — audio is a small premium that's wasted in that flow; drop to [`veo3.1-fast`](/models/veo3-1-fast) without audio.

## Next

* [`veo3.1-fast`](/models/veo3-1-fast) — cheaper tier
* [Multimodal endpoints](/concepts/multimodal) — overview