Private AI

Nano GPT logo
NanoGPT
Nano GPT logo
NanoGPT
Nano GPT logo
NanoGPT
Search
 
 
 
 
All conversations

Create

Chat
Studio

Explore

Models
Pricing
AI Apps
Blog

Library

Gallery
Workspace
Usage
API
Suggestions and bugs
Terms

|

Privacy

|

Refunds
Balance
Subscription
Teams
Referrals
Presets
Settings
FAQPress & mediaBrand assetsBenchmarksEarn crypto
  1. Models
  2. Image
  3. Luma

Luma UNI-1

Try Model

Luma's multimodal image model for high-fidelity text-to-image generation and prompt-guided image edits in one model. Supports reference images, optional web grounding, manga style, and PNG/JPEG outputs.

Added Jun 9, 2026

Details

Approx. Price
$0.042 per image
Model type
both

Settings

Generation controls available for this model.

Images Per Run

Fixed 1

Output images

Input Images

Up to 9

Reference/edit images accepted • Route max 30 MB

Output Sizes

Square
Portrait
Landscape

Number of Images

Default

1

Output Format

Select

Default

png

Options (2)

PNG, JPEG

Encoding format

Reference Images

Text

Default

N/A

Optional JSON array of reference image URLs

Resolution

Default

16:9

Options (9)

3:1 (3:1), 2:1 (2:1), 16:9 (Landscape (16:9)), 3:2 (Landscape (3:2)) +5 more

Style

Select

Default

auto

Options (2)

Auto, Manga

Visual style

Web Search

Toggle

Default

No

Allow web-sourced visual grounding for text-to-image requests

Benchmarks

Human preference benchmarks sourced from Artificial Analysis.

Text to Image

#51 / 106

ELO

992.0

Appearances

5,559

95% CI

-12/12

Image Editing

#40 / 77

ELO

1043.0

Appearances

9,459

95% CI

-13/13

Release Date 2026-05 · Matched as Luma UNI 1

Artificial Analysis API

Examples

Loading examples…

Related image models

Compare Luma UNI-1 with similar models from the same provider or model family.

Luma UNI-1 Max

luma/agent/uni-1/v1/max

Maximum-fidelity UNI-1 generation and editing for richer detail, stronger prompt adherence, reference following, and high-quality final stills.

Qwen Edit Uncensored

qwen-edit-uncensored

Uncensored Qwen image editing. Edit an image or combine up to three references with a text instruction. Supports native 1K output, flexible aspect ratios, and PNG, JPEG or WebP format. Requires an input image; edit prompts can contain up to 1,500 characters.

Qwen Image 3 Uncensored

qwen-image-3-uncensored

Uncensored Qwen Image 3 text-to-image. Create images from prompts up to 10,000 characters, with native 1K or 2K output, eight aspect ratios, and PNG, JPEG or WebP format. $0.04 per image at either resolution.

Ideogram Vector

ideogram/vector

Ideogram Vector creates downloadable SVG graphics from text, with up to five optional reference images. Generate 1–8 SVGs with low, medium, or high quality, 18 aspect ratio choices including Auto, and an optional seed. Switch to image conversion to turn one JPEG, PNG, or WebP into SVG without a prompt; choose Default for finer detail or Simple for fewer shapes. References allow 10 MB each; conversion accepts up to 50 MB.

FIBO 2

bria/fibo-2

FIBO 2 by Bria creates images from text and edits up to four ordered source and reference images. Use Inspiration to create a new scene from one visual reference, or a same-size mask to edit one image while preserving pixels outside the mask. Generation and multi-image edits support 21 aspect ratios at about 1MP; single-image edits preserve the source aspect ratio. Supports four photographic styles, reproducible seeds, and reusable structured VGL prompts.

MOSS Image

fal-ai/moss-image

MOSS Image by OpenMOSS generates images from natural-language prompts, tag sequences, or serialized JSON layouts, with Chinese and English text rendering and control over text placement and object arrangement. Create 1–4 images per request in JPEG, PNG, or WebP, with dimensions from 512 to 2048 pixels per side aligned to multiples of 16. Optional prompt expansion plans a structured layout and falls back to your original prompt if unavailable. Supports negative prompts, 1–100 steps, guidance from 0 to 10, and reproducible seeds. Text-to-image only.