# Ray 3.2

Ray 3.2 (Ray3.2) is the newest Luma Ray 3 model, announced on 9 June 2026 for studios, agencies and marketing teams who want control over the shot. It makes 5 or 10 second clips without sound. In Sloosh, Ray 3.2 takes text and optional start and end frames, at up to 1080p.

Web page: https://sloosh.ai/models/ray-3.2

[Use Ray 3.2](https://sloosh.ai/studio/video?video-model=ray-3.2)

## Key facts in Sloosh

|   |   |
| --- | --- |
| Made by | Luma |
| Makes | video (with sound: never) |
| Inputs | text; start frame (optional); end frame (optional); no reference images or videos |
| Length | 5 or 10 seconds |
| Resolution | 360p, 720p (default), 1080p |
| Aspect ratios | Auto, 9:16, 3:4, 1:1, 4:3, 16:9, 21:9 |
| Prompt limit | 6,000 characters |

## Best for

- Clips you will score or voice in post: Luma says text2video and image2video don't generate audio, so you add voiceover, music and effects separately ([Luma Ray page](https://lumalabs.ai/ray)).
- Animating between a start and end frame: Luma's API takes either, both or neither as anchor images ([Luma video generation guide](https://docs.agents.lumalabs.ai/guides/videos/generation/)).
- Brand and campaign work: Luma built it with creatives from entertainment, advertising and gaming, for studios, agencies and enterprise marketing teams ([Luma announcement](https://lumalabs.ai/news/introducing-ray-3-2)).
- Cheap exploration at 360p before a 1080p final: Luma labels 360p as its draft tier ([Luma Ray page](https://lumalabs.ai/ray)).

## How to use Ray 3.2 in Sloosh

1. Open [Video](https://sloosh.ai/studio/video?video-model=ray-3.2) and pick Ray 3.2 in the model chip.
2. Optionally add a start frame, an end frame, or both.
3. Write the prompt: subject, action, setting, camera, lighting, style.
4. Pick 5 or 10 seconds, the resolution and aspect ratio.
5. Check the price on the Generate button, then generate.

## Ray 3.2 prompting tips

- Use Luma's framework: "Subject + Action + Setting + Camera + Lighting + Style", with the key details in the first 20 to 30 words ([Luma: how to write AI video prompts](https://lumalabs.ai/news/write-ai-video-prompts)).
- Aim for about 100 to 150 words ([Luma: how to write AI video prompts](https://lumalabs.ai/news/write-ai-video-prompts)).
- Use only one camera movement per clip, described with its pace, for example "gentle, steady push-in over 5 seconds"; add "static camera" if movement causes problems ([Luma: how to write AI video prompts](https://lumalabs.ai/news/write-ai-video-prompts)).
- Phrase what you want positively instead of listing what to avoid, and add text or logos in post rather than asking for them ([Luma: how to write AI video prompts](https://lumalabs.ai/news/write-ai-video-prompts)).
- Be specific about subject, motion, camera movement, lighting and pacing ([Luma video generation guide](https://docs.agents.lumalabs.ai/guides/videos/generation/)).

## Ray 3.2 example prompts

**Prompt 1**

```text
A barista pours steamed milk into a latte, forming a leaf pattern, in a sunlit corner cafe. Gentle, steady push-in over 5 seconds toward the cup. Warm morning light through the window, shallow depth of field, natural film look.
```

**Prompt 2**

```text
A red vintage convertible drives along a coastal cliff road at sunset, its scarf-wearing driver's hair moving in the wind. Tracking shot from the side at matching speed. Golden backlight, long shadows, cinematic colour.
```

**Prompt 3**

```text
From the start frame of a folded paper crane on a desk to the end frame of the crane in flight above it: the crane lifts off and flaps upward. Static camera. Soft daylight, clean white background.
```

## How it compares in Sloosh

|   | Ray 3.2 | MiniMax H3 Max | Wan 3 |
| --- | --- | --- | --- |
| Inputs | text, start/end frame | text, start/end frame | text, frames or references |
| Length | 5, 10 s | 5 to 15 s | 5, 10, 15, 30 s |
| Max resolution | 1080p | 1080p | 1080p |
| Sound | never | always | optional |
| Prompt limit | 6,000 | 50,000 | 20,000 |

## Where else it runs

Luma's own app and API ([Luma Ray page](https://lumalabs.ai/ray), [Luma announcement](https://lumalabs.ai/news/introducing-ray-3-2)).

## Frequently asked questions

### When was Ray 3.2 released?

Luma's announcement page is dated 9 June 2026.

### Does Ray 3.2 make sound?

No. Luma's text-to-video and image-to-video don't generate audio; add sound in your editor.

### How long are Ray 3.2 videos in Sloosh?

5 or 10 seconds.

### What resolution does Ray 3.2 make?

360p, 720p or 1080p in Sloosh.

### Can Ray 3.2 use start and end frames?

Yes, both are optional in Sloosh.

### What is the difference between Luma Ray 3 and Ray 3.2?

Luma calls Ray3.2 a significant update to the Ray3 family and "our newest and most versatile video model".

## Related

- Models: [MiniMax H3 Max](https://sloosh.ai/models/minimax-h3-max), [Wan 3](https://sloosh.ai/models/wan-3.0), [Seedance 2.0 Fast](https://sloosh.ai/models/seedance-2.0-fast)

## Sources

- [Luma: Introducing Ray3.2](https://lumalabs.ai/news/introducing-ray-3-2)
- [Luma: Ray page](https://lumalabs.ai/ray)
- [Luma: video generation guide](https://docs.agents.lumalabs.ai/guides/videos/generation/)
- [Luma: How to write AI video prompts](https://lumalabs.ai/news/write-ai-video-prompts)