Skip to content
Radiant CanvasRadiant Bear Games Mac App Store
New in 1.30New design with floating glass panels

AI image generator for Mac

Make and edit images on your Mac.

Create images from a prompt, change a photo by describing the edit, and upscale to 4096 pixels. Eight image models run on Apple silicon. No account, no uploads and no credits.

Free download. The PRO trial is for eligible customers on the monthly plan. Needs an Apple silicon Mac with macOS 26.2 or later. Requirements

Radiant Canvas Create workspace with a generated image of a greyhound at a computer, the prompt panel on the left and Options on the right

8 image modelsPlus the SeedVR2 upscaler

Edit with wordsFree with Qwen and FLUX.2 Klein

Up to 4096 pixelsFree upscaling

Stays on your MacWorks offline after download

What you can do

One app for making, editing and keeping images.

Create with the model picker open and three LoRAs set in the LoRAs panel

Create

Describe an image and generate it. Pick a model, a shape and a size, then make variations or queue up to 100 images.

  • Run one prompt on several models and compare the results side by side.
  • Ming makes designs with text and transparent backgrounds.
  • With PRO, add up to ten of your own LoRA or LoKr adapters and save them as presets.
Compare models on Mac
Before and after view of an edit that replaced a cat with a dog in the same room

Edit

Add a photo, say what to change and generate. Shown here: “Replace the cat with a dog”, made with FLUX.2 Klein 4B.

  • FLUX.2 Klein uses one reference image. Qwen Image Edit 2511 uses up to three.
  • Check the result against the original in the Before / After view.
  • Editing is free. AI edits can change details you meant to keep, so check each result.
How image editing works
A saved image open in the Gallery viewer, with its details on the right

Upscale

Add an image and choose an output size. SeedVR2 3B enlarges it and restores fine detail. There is no prompt to write.

15362048256030724096
  • Raise Softness for small, noisy or compressed images.
  • Upscale any saved image from the Gallery or the image viewer.
  • A 7.3 GB download. Macs with 16 GB of memory can upscale up to 3072 pixels.
Upscaling with SeedVR2
Studio with an image-to-image workflow: input image, model, prompt, Blur / Sharpen, histogram and previews, with the Inspector open

Studio

Connect prompts, models, images and tools on a board. Check the run plan, then save the workflow to use again.

  • Basic text-to-image workflows are free.
  • PRO adds nodes for image-to-image, editing, masks, depth guidance, batches, model and seed studies, upscaling and export.
  • Start from a sample, your Create settings or an image in the Gallery.
Try PRO free for 3 days

Compare modelsRun one prompt on several installed models at once.

Your own LoRAsUp to ten LoRA or LoKr adapters with PRO.

MCP ServerMake images from Claude, Codex, Cursor and other AI apps on this Mac.

Demo modeTry the app before you download a model.

Model Manager

Eight models. Pick the ones that fit your Mac.

Each model is a separate download. Most come in a standard and a higher-precision version, 14 options in all. Model Manager shows the size and memory needs before you download, and you can keep several installed.

Models, what they are good at, download sizes and memory needs
ModelGood forDownloadMemory
FLUX.2 Klein 4B4-bit · 8-bitFast images and editing with one reference. The smallest download.4.7 GB
8.6 GB
16 GB minimum
24 GB minimum
Qwen Image Edit 25114-bit · 8-bitImages and editing with up to three references.19.9 GB
31.8 GB
16 GB minimum, 32 GB recommended
64 GB minimum
Krea 2 TurboINT4 ConvRot · Q8Detailed images. INT4 is the pick for Macs with less memory.15.5 GB
22.2 GB
Best for low memory
16 GB minimum
Z-Image Turbo8-bit · BF16Realistic images, accurate prompts and clear text.13.3 GB
20.5 GB
16 GB recommended
32 GB minimum
ERNIE-Image Turbo4-bit · 8-bitDetailed layouts and text in many languages.6.7 GB
12.4 GB
16 GB recommended
32 GB minimum
Ming-Image-0.1-Design4-bit · 8-bitDesigns, lettering and transparent backgrounds.13.4 GB
25.1 GB
16 GB minimum
MOSS-Image SFT Turbo8-bitLong prompts, tag lists and JSON layouts. Clear text in English and Chinese.16.1 GB16 GB minimum
Microsoft Lens TurboMLX 8-bitFast four-step images.16.8 GB16 GB minimum
SeedVR2 3BUpscalerEnlarging an image up to 4096 pixels.7.3 GB16 GB minimum, 24 GB for 4096 px
Model Manager suggesting 8-bit for this Mac, with Standard INT4 ConvRot at 15.5 GB and Higher precision 8-bit at 22.2 GB

Standard or higher precision?

Standard versions are smaller and use less memory. Higher precision uses more of both and can be slower. It changes small details, but it does not promise a better image. Model Manager suggests one for your Mac.

Model licenses differ, and each one is shown before its download.

How long one image takes

1024 × 1024 text-to-image times on a MacBook Pro with M5 Max and 48 GB. Fast mode needs PRO. Times vary with your Mac, the settings, the image size and LoRAs. Qwen, MOSS-Image and the higher-precision versions were not measured.

FLUX.2 Klein 4B
Fast: 3.1sStandard: 6.2s
Microsoft Lens
Fast: 3.4sStandard: 4.0s
ERNIE-Image
Fast: 7.5sStandard: 25.4s
Z-Image
Fast: 11.2sStandard: 20.8s
Krea 2 INT4
Fast: 15.1sStandard: 20.2s
Krea 2 Q8
Fast: 20.5sStandard: 26.8s
Ming 4-bit
Fast: 15.2sStandard: 27.2s

Private by design

Your prompts and images stay on your Mac.

Runs on Apple silicon

Every model runs on your Mac. Nothing is sent to a cloud service to make or edit an image.

Works offline

After a model is downloaded, you can create and edit with no internet connection.

No account or tracking

No sign-in, ads, analytics or credits. Support only sees what you put in an email.

Labeled as AI

Each image carries a hidden label and Content Credentials that say it was made with AI, as EU law requires. They hold no prompt or personal data.

Read the privacy policy

Made in Radiant Canvas

A few images made on a Mac.

Each caption names the model that made the image.

Translucent cosmic creature feeding on a dying star
Z-Image Turbo 8-bit
Futuristic city beneath a transparent Antarctic ice shelf
FLUX.2 Klein 4B 4-bit
Armored traveler among enormous stone faces in a forest
Krea 2 Turbo INT4 ConvRot
White palace rising from a moonlit ocean with flying whales
ERNIE-Image Turbo 4-bit
Steam train crossing a bridge above a sea of clouds
Krea 2 Turbo Q8
Medieval city built across the backs of stone giants
Z-Image Turbo 8-bit

Free and PRO

Free to start. PRO when you want more.

Every model is free to download and use. PRO adds more image shapes, more control and the advanced Studio nodes. Eligible customers can try monthly PRO free for 3 days.

Free and PRO features
FeatureFreePRO
All eight models and model comparisonIncludedIncluded
Square images, variations, the queue and the GalleryIncludedIncluded
Editing with Qwen and FLUX.2 Klein in EditIncludedIncluded
Upscaling with SeedVR2 in UpscaleIncludedIncluded
Basic Studio text-to-image workflowsIncludedIncluded
4:3, 9:16 and 16:9 shapes on supported modelsNot includedIncluded
Manual controls and negative promptsNot includedIncluded
Fast mode, ML Enhance and PNG exportNot includedIncluded
Up to ten LoRA or LoKr adaptersNot includedIncluded
Advanced Studio nodes for editing, masks, batches, upscaling and exportNot includedIncluded

PRO is monthly, yearly or a one-time lifetime purchase, with the same features on every plan. The monthly trial renews as a paid subscription unless you cancel. The app shows current prices and whether you can get the trial before you buy.

Requirements

Which Macs can run it?

An Apple silicon Mac (M1 or later) with macOS 26.2 or later. How much memory you need depends on the model; see the model list.

Intel Macs are not supported

The models run on Apple silicon only.

Memory matters most

Most models need 16 GB. On an 8 GB Mac, Model Manager suggests Krea 2 INT4. Higher-precision versions need 24 to 64 GB.

Downloads are large

From 4.7 GB to 31.8 GB per model. Downloads can pause and resume, and files are checked before install.

Models stay in the app

They can't be moved to an external drive yet. Delete any model and download it again later.

Apple silicon and memory guide

Questions

Questions about Radiant Canvas

Can I edit an existing image with AI on my Mac?

Yes. In Edit, select FLUX.2 Klein or Qwen Image Edit 2511, add a reference image, and describe what you want to change. Klein accepts one reference image; Qwen accepts up to three. Editing runs locally and is included free within the free image-size options. Review each result, since AI editing can also change details you wanted to keep.

What is the difference between Qwen Image Edit 2511 and FLUX.2 Klein?

Both generate images from text and edit existing images locally. FLUX.2 Klein 4B uses one reference image and has a smaller download of about 4.7 GB. Qwen Image Edit 2511 uses up to three references and has a 4-bit download of about 19.9 GB. These standard variants require at least 16 GB of unified memory; Qwen recommends 32 GB. Higher-precision 8-bit variants are also available: Klein needs at least 24 GB and Qwen needs 64 GB. PRO adds Qwen negative prompts and Fast mode for larger images.

Is Qwen and FLUX.2 Klein image editing free?

Yes. Both models and reference image editing in Edit are included in the free app, within free square image sizes. Running the Image Edit node in Studio requires PRO. There are no per-image charges or cloud credits.

Can Radiant Canvas upscale images on my Mac?

Yes. Open Upscale, add an image, and choose an output size from 1536 to 4096 pixels on the long edge. SeedVR2 3B enlarges the image and restores fine detail without a prompt. It is a 7.3 GB download and needs at least 16 GB of unified memory; Macs with 16 GB can upscale up to 3072 pixels. Upscaling in Upscale is free. The SeedVR2 Upscale node in Studio requires PRO.

Does Radiant Canvas upload my prompts or images?

No. Prompts, reference images, image generation and editing, results, Gallery data, settings, and imported LoRAs stay on your Mac. Model downloads come from Hugging Face or a mirror. Apple handles App Store purchases and the optional ML Enhance model. The app has no analytics, telemetry, tracking, account, or developer backend. This website uses optional consent-based analytics separately.

Does Radiant Canvas work on M1, M2, M3, M4, and M5 Macs?

Yes, if the Mac runs macOS 26.2 or later and has enough unified memory for the selected model. Intel Macs are not supported. Performance varies by chip, memory, model, image size, and settings.

How much unified memory do I need?

Ming 4-bit and 8-bit, MOSS-Image, Krea 2 Q8, FLUX.2 Klein 4B, Microsoft Lens Turbo, and Qwen Image Edit in their standard variants require at least 16 GB. Qwen recommends 32 GB. Z-Image and ERNIE-Image recommend 16 GB. Krea 2 INT4 is the app’s recommended lower-memory option. Higher-precision FLUX.2 needs at least 24 GB, Z-Image BF16 and ERNIE 8-bit need 32 GB, and Qwen 8-bit needs 64 GB. Model Manager shows guidance before download.

How much storage does each model use?

Approximate downloads range from 4.7 GB for FLUX.2 Klein 4-bit to 31.8 GB for Qwen Image Edit 8-bit. The model list above shows every option. The app checks the remaining space needed before it starts or resumes a download.

Can model files be stored on an external drive?

Not currently. Model files are stored in Radiant Canvas app storage. You can delete models separately in Model Manager and download them again later without deleting Gallery images.

Can Radiant Canvas work offline?

Yes. Generation and reference image editing work offline after your chosen models are installed. New model downloads, App Store purchases and restoration, and the first download of Apple’s optional ML Enhance model require a connection.

Is Radiant Canvas a Stable Diffusion app?

No. Its eight models are FLUX.2 Klein, Qwen Image Edit, Krea 2, Z-Image, ERNIE-Image, Ming-Image-0.1-Design, MOSS-Image, and Microsoft Lens. Six of them come in a standard and a higher-precision version, which makes 14 download options. SeedVR2 3B is a separate upscaler.

Can I use generated images commercially?

Model licenses differ. Radiant Canvas shows the relevant license before each download. Review that license and any adapter license for your intended use. The app does not make one blanket commercial-use guarantee for every model and LoRA.

What is the difference between Krea 2 Q8 and INT4 ConvRot?

Q8 is the larger, higher-precision option and requires at least 16 GB of unified memory. INT4 ConvRot is smaller and is the app’s recommended Krea option for lower-memory Macs.

Does Radiant Canvas support custom LoRAs and LoKr adapters?

Yes. PRO can add up to ten compatible LoRA or LoKr adapters and save presets for supported model families. Ming does not support adapters.

What can I build in Studio?

Studio is a visual workspace for repeatable local image workflows. Connect typed nodes, preview images, inspect the run plan, and save workflows for later. Basic text-to-image workflows are free. PRO adds advanced nodes for image input and editing, masks, studies, batches, ControlNet guidance, enhancement, metadata, and export.

Do I need to pay to generate images?

No. The free app includes all eight models, square generation, reference image editing in Edit, SeedVR2 upscaling, one-prompt model comparison, up to 100 variations, a queue of up to 100 outputs, prompt history, metadata, watermarks, and the local Gallery. It also includes basic Studio text-to-image workflows.

Radiant Canvas app icon

Free on the Mac App Store

Make your first image.

Download the app, pick a model and start. Everything runs on your Mac.

Download free on theMac App Store