A humpback whale floating over a wheat field at dusk, watched by a figure in a red coat

Feature

AI image generator

Write a prompt or upload references and generate images from the latest versions of Seedream, GPT Image, Nano Banana, Ideogram, and other leading models, side by side on the same canvas, all on one subscription.

See what's been made

A TIDE coconut water can on ice between coconut halves and limes
A Malibu After Hours risograph poster of hand-drawn flowers and dancing figures
A risograph print of a black cat on a windowsill under a teal moon
A man in a flowing blue robe standing against a pink wall
A close-up of gold hoop and pearl earrings on an ear

From prompt to finished image

  1. Start with a prompt or referencesDescribe the image you want and attach references in one message.
  2. Brief your agentTell Mel, the Melius creative execution agent, or your own agent what you'd like to make. It sets up the nodes, choosing the model to run for each one, then generates your image.
  3. Compare and refineRun one prompt through two models and keep the result you prefer. If one step misses, edit that node and rerun it without redoing the rest. Then upscale the image you keep.

What you can make

Every model has its strengths. Melius lets you use each one for the part it does best.

Burrata with roasted cherry tomatoes and olive oil, a hand reaching in with bread

Photorealistic images

Portraits, lifestyle scenes and editorial shots. Nano Banana Pro follows a style brief closely and is great with fabric, food and skin tone.

A Night Shift jazz poster in bold black type over blue and orange circles

Images with readable text

Posters and packaging where the words have to be right. Ideogram 4.5 is built around clean type, and its edits change only what you ask for.

Reference: a model in a white linen shirt
Reference: a white mascara tube with its wand
The result: the same model applying the mascara

Images built from your references

Connect up to 15 reference images to a GPT Image 2.5 node to generate or edit from them.

An infographic on how coffee gets to your cup: six steps from harvest to brewing, a caffeine chart and a production pie chart

Layout-heavy designs

Infographics and UI screens with a lot to fit in. Qwen Image 3 builds a dense layout in a single pass and keeps text as small as 10 pixels readable.

Commuters, flower carts and pigeons crossing a sunlit train station concourse

Scenes with a lot going on

Seedream 5.0 Pro reads where each element sits in the frame, so things land where your prompt puts them.

The same steel water bottle in eight colorways

Quick drafts and ideas

Nano Banana 2.1 generates quickly, so you can try a lot of directions early and save the slower models for the one you pick.

Try out Melius

Frequently Asked Questions

What is the best AI image generator?
Nano Banana Pro is the model to pick for photorealism, Ideogram 4.5 for typography, GPT Image 2.5 for working from reference images, Qwen Image 3 for layout-heavy designs and Seedream 5.0 Pro for scenes with a lot of elements to place. Melius runs all of them, so you can test them on the same prompt before you commit.
How is Melius different from Midjourney?
Midjourney runs its own image models in its own app. On Melius you get Seedream, GPT Image, Nano Banana, Ideogram and other leading models in one place, with Mel choosing a model for each step and every prompt left open for you to edit.
Can you use Seedream, GPT Image and Nano Banana without separate subscriptions?
Yes. One Melius subscription gives you Seedream, GPT Image, Nano Banana and the other image models, so you don't need separate accounts with ByteDance, OpenAI or Google. The current lineup is on the models page and plans are on the pricing page.
Can you generate an AI image from an existing image?
Yes. Upload a photo, sketch or other type of image to Melius and tell Mel what you want to create.
Can you use Melius for text-to-image generation?
Yes. Type a description into an image node and run it through any of the image models on Melius, such as Nano Banana Pro, Seedream 5.0 Pro or Ideogram 4.5.
Can you generate AI images through an API or MCP?
Yes. Melius has an API, a CLI and an MCP (Model Context Protocol) server, so you can generate images from your own code or have an agent such as Claude do it for you. Setup is on the MCP, API and CLI page.

Related pages

Contact Sales

Tell us about your team and what you’re making, and we’ll get back to you with how Melius fits.