# Uni-1

Uni-1 is Luma's first unified understanding and generation model: one model that reasons about a prompt and then draws, instead of two systems chained together. Luma highlights reference-guided generation with up to nine references, directable editing, text on signs and labels, and spatial reasoning. In Sloosh it makes 2K images from a prompt and up to 9 references.

Web page: https://sloosh.ai/models/uni-1

[Use Uni-1](https://sloosh.ai/studio/image?image-model=uni-1)

## Key facts in Sloosh

|   |   |
| --- | --- |
| Made by | Luma |
| Makes | image |
| Inputs | text, up to 9 reference images |
| Resolution | 2K |
| Aspect ratios | Auto, 16:9, 3:2, 1:1, 2:3, 9:16 (default Auto) |
| Prompt limit | 6,000 characters |

## Best for

- Reference-guided images: "Pass up to nine reference inputs per request to preserve identity, composition, style, or any combination of them" ([Luma](https://lumalabs.ai/news/uni-1-1-api)).
- Text on signs, labels and book covers, and spatial logic for shadows, perspective and layout ([Luma API docs](https://docs.agents.lumalabs.ai/guides/model/)).
- Named cultural styles such as manga screentone, ukiyo-e and film noir ([Luma API docs](https://docs.agents.lumalabs.ai/guides/model/)).
- Storyboards and keyframes that stay consistent, which Luma lists as a capability ([Luma](https://lumalabs.ai/uni-1)).

## How to use Uni-1 in Sloosh

1. Open [Uni-1 in Studio](https://sloosh.ai/studio/image?image-model=uni-1).
2. Add up to 9 references and give each a role in the prompt.
3. Describe the image with concrete terms (lens, light, named style).
4. Leave the ratio on Auto or pick one. Output is 2K.
5. Press Generate; the price shows first.

## Uni-1 prompting tips

- "References only work if you tell the model what they are for." Use "Use IMAGE1 (description) as a [ROLE] reference", one role per image ([Luma start guide](https://lumalabs.ai/learning-hub/luma-uni-1-model)).
- "Put exact text in quotes in your prompt" ([Luma API docs](https://docs.agents.lumalabs.ai/guides/model/)).
- Prefer concrete technical terms ("Golden hour, 85mm lens, shallow depth of field") and named aesthetics over words like "beautiful" ([Luma start guide](https://lumalabs.ai/learning-hub/luma-uni-1-model)).
- When editing, state what changes and what must stay untouched ([Luma start guide](https://lumalabs.ai/learning-hub/luma-uni-1-model)).
- Luma suggests about 80 to 250 words for text-to-image and 100 to 300 with references ([Luma start guide](https://lumalabs.ai/learning-hub/luma-uni-1-model)).

## Uni-1 example prompts

**Prompt 1**

```text
Use IMAGE1 (a woman in a red wool coat) as a character reference and IMAGE2 (a rainy street painting) as a style reference. Show the woman waiting at a tram stop in the rain at dusk, golden streetlight, 50mm lens, shallow depth of field.
```

**Prompt 2**

```text
A paperback book cover in ukiyo-e woodblock style: a lone fisherman in a small boat under a huge wave, the title "The Long Tide" in bold serif letters at the top.
```

**Prompt 3**

```text
A four-panel storyboard, same girl and same yellow umbrella in every panel: she leaves her house, walks through the rain, meets a stray dog, shares the umbrella with it.
```

## How it compares in Sloosh

| Model | Resolution | References | Prompt limit |
| --- | --- | --- | --- |
| Uni-1 | 2K | 9 | 6,000 |
| [Uni-1 Max](https://sloosh.ai/models/uni-1-max) | 2K | 9 | 6,000 |
| [Nano Banana 2](https://sloosh.ai/models/gemini-3.1-flash-image-preview) | 512 to 4K | 14 | 50,000 |
| [Krea 3](https://sloosh.ai/models/krea-3) | 1K, 2K | 8 | 5,000 |

Uni-1 is the exploring tier. Uni-1 Max takes the same inputs and settings; Luma says it produces higher-quality output at a higher per-image price.

## Where else it runs

Luma's own app and the Luma API (Uni-1.1 API opened 5 May 2026, per [Luma](https://lumalabs.ai/news/uni-1-1-api)).

## Frequently asked questions

### What is Luma Uni-1?

Luma's "first unified understanding and generation model" ([Luma](https://lumalabs.ai/uni-1)), one model that reasons about the prompt and draws, instead of separate systems stitched together.

### How many reference images does Uni-1 take?

Up to 9, in Sloosh and per Luma.

### What resolution does Uni-1 make?

2K in Sloosh; Luma prices text-to-image at 2048px ([Luma](https://lumalabs.ai/uni-1)).

### Uni-1 vs Uni-1 Max?

Luma says the two share the same parameter set and that Max gives higher-quality output at a higher price. Use Uni-1 to explore and [Uni-1 Max](https://sloosh.ai/models/uni-1-max) for finals.

### Can Uni-1 write text in images?

Yes. Luma lists text rendering on signs, labels and book covers, including non-Latin scripts such as Chinese, Japanese and Arabic.

## Related

- Models: [Uni-1 Max](https://sloosh.ai/models/uni-1-max), [Nano Banana 2](https://sloosh.ai/models/gemini-3.1-flash-image-preview), [Krea 3](https://sloosh.ai/models/krea-3)
- Answers: [What is the best AI image model in 2026?](https://geo.sloosh.ai/posts/what-is-the-best-ai-image-model-in-2026)

## Sources

- [Luma: Uni-1](https://lumalabs.ai/uni-1)
- [Luma: Introducing the Uni-1.1 API](https://lumalabs.ai/news/uni-1-1-api)
- [Luma API docs: Models](https://docs.agents.lumalabs.ai/guides/model/)
- [Luma: Uni-1 start guide](https://lumalabs.ai/learning-hub/luma-uni-1-model)