Arumi

AI image · Text to image and editing

AI Image Generator

Generate images from text or reference images, then edit and regenerate the result.

  • 18 image models
  • Up to 4K
Loading models…
Sign in to generate

Examples and their prompts

Results from this generator. A prompt can be used as it is, or copied and edited.

Available image models

Every model shows only the settings it supports. Select a model in the box above; switching models keeps the prompt.

Model list updated

ModelSuited toResolutionInputsCurrent price

Seedream 4.0

BytePlus		Unified 4K generation and editing1K–4KUp to 14 images

Seedream 4.5

BytePlus		Consistent multi-image editing with typography2K–4KUp to 14 images

Seedream 5.0 Flash

BytePlus		Fast low-cost image generation1K–2KUp to 10 images

Seedream 5.0 Flash Layerize

BytePlus		Image split into editable layers1K–2KUp to 1 image

Seedream 5.0 Pro

BytePlus		Reasoning-driven infographic generation1K–2KUp to 10 images

Seedream 5.0 Pro Layerize

BytePlus		High-fidelity layered image breakdown1K–2KUp to 1 image

Nano Banana Pro

Google		Studio-quality 4K visuals with text rendering1K–4KUp to 14 images

Nano Banana 2

Google		High-efficiency generation and editing0.5K–4KUp to 14 images

Nano Banana 2 Lite

Google		Rapid 1K image ideation1KUp to 14 images

Nano Banana 2.1

Google		4K images with accurate text layouts1K–4KUp to 14 images

GPT Image 1.5

OpenAI		Improved prompt adherence for image generation—Up to 16 images

GPT Image 2

OpenAI		Prompt-faithful image text rendering1K–4KUp to 16 images

GPT Image 2.5 Flare

OpenAI		Fast everyday image generation1K–4KUp to 16 images

GPT Image 2.5 Sunburst

OpenAI		Precision-focused image editing1K–4KUp to 16 images

Grok Imagine Image 2.0

xAI		Multi-output image variations1K–2KUp to 3 images

Qwen Image 3.0

Alibaba		Faster dense-layout image generation1K–2KUp to 3 images

Qwen Image 3.0 Pro

Alibaba		Small multilingual text rendering1K–2KUp to 3 images

Wan 2.5 Image

Alibaba		Enhanced aesthetics and realistic texture—Up to 2 images

How to generate an image

Three steps from a description to an image.

  1. Describe the image

    Choose a model from the picker at the top of the form.

  2. Adjust the settings

    Write a prompt; add reference images if the model supports them.

  3. Generate

    Set size and quality, check the credit price, and click Generate.

Using reference images

A reference image shows the model what is meant: the same subject, a style or a composition.

  1. Original photo
  2. Reference
  3. Result

Frequently asked questions

Answers to frequently asked questions.

Which model should I choose?

Hover over a model in the picker to see what it is suited to and which inputs it accepts. To compare, start with the default model and then try a second model on the same prompt; the prompt is kept when switching models.

Can I edit my own photo?

Yes, with any model that accepts reference images: upload the photo and describe the change.

What happens if a generation fails?

Failures caused by us or blocked by our content checks are free. If the model service does not charge for a failure, neither do we. If it does, you pay for the actual usage, capped at the quote, and the record shows the charge.

Need more information? Email us at support@arumi.ai.

Start generating

Create images, video, audio and music from a text description with Arumi.