Skip to content

CircleStone Labs · 2B · NVIDIA Cosmos-Predict2 fine-tune

Run Anima on iPhone, iPad, and Mac: Anime-Style Images

Anima makes anime: characters, scenes, and styles, plus other illustration that is not meant to look like a photograph, written as the comma-separated tags anime image boards use. CircleStone Labs left out photorealism on purpose, and it runs on every iPhone, iPad, and Mac the app supports.

Parameters
2B
License
CircleStone Labs Commercial License (NVIDIA Open Model License flow-through)
Released
January 2026
Download
2.51 GB

View on Hugging Face

Anima runs in Private Diffusion, a local AI image generator for iPhone, iPad, and Mac. Every model in the app is trained with quantization aware distillation (QAD) to stay near-lossless at a fraction of its full size. Download it once inside the app.

Pre-order on the App Store
Anima in Private Diffusion Studio on iPhone, with a finished image

Anima Requirements: Will It Run on My Device?

  • iPhone

    Runs on every supported iPhone

    iPhone 15 Pro and later, from 8 GB of memory up.

  • iPad

    Runs on every supported iPad

    Any iPad with an M-series chip.

  • Mac

    Runs on any Apple Silicon Mac

    M1 or later, from 8 GB of memory up.

Check your device

Anima Specs

Download

Download Size
2.51 GB

Model

Maker
CircleStone Labs
Based On
NVIDIA Cosmos-Predict2 fine-tune
Parameters
2B
Quantization
Trained INT6 (QAD)
License
CircleStone Labs Commercial License (NVIDIA Open Model License flow-through)Your images are yours, including for commercial use, except that you may not use them to train a model that competes with Anima.

Generation

Steps
8-12 (adjustable)
Upscaler
Off · 2x
Negative Prompt
No
Rewrite Prompt
No
Output Sizes
  1. S · 512 pxEvery device
  2. M · 768 pxEvery device
  3. L · 1024 pxEvery device
  4. XL · 1536 px16 GB

Anima Examples and Prompts

55 of 55 prompts

Browse Pixel Art prompts

What Anima Is Good At

It knows a lot of anime. CircleStone Labs trained it on several million anime images, with anime data running to September 2025, plus about 800,000 other artworks, so it recognizes many characters and series by name. A steps slider runs from 8 to 12: more steps take a little longer and add a little detail.

Built on NVIDIA Cosmos. NVIDIA Open Model License

Pick It For

  • Anime characters, scenes, and styles, prompted with tags such as "1girl, solo, cherry blossoms"
  • Illustration and painted artwork beyond anime, as long as it is not meant to pass for a photo
  • A 2.51 GB download that runs on an 8 GB iPhone

Know Before You Download

  • No negative prompt and no on-device prompt rewriting
  • Text in the image holds up for a word or a short phrase, not a sentence
  • Upscaling stops at 2x
  • Its license allows commercial use of your images, but not training a competing model on them

Anima in Private Diffusion vs. Draw Things vs. ComfyUI

Anima is the same model wherever you run it. The apps differ in how they shrink it to fit, which chips do the work, and who sees your prompts.

Private Diffusion compared with Draw Things and ComfyUI
FeaturePrivate DiffusionDraw ThingsComfyUI
Image QualityNear-lossless: every model is quantization aware distilled (QAD) against the full-size originalShrinks models after training (post-training palettization)Runs the weights you load
SpeedGPU and Neural Engine together, plus Neural Accelerators on M5, M6, A19 Pro, and A20 ProGPU; the Neural Engine only for i8x models on M3 and M4PyTorch MPS, the slowest inference framework on Apple platforms
PrivacyCollects no dataPrivacy label lists Purchases, Identifiers, and User ContentSends telemetry to PostHog and Datadog, checked by default on first launch
SecurityApp Sandbox and Hardened RuntimeApp Sandbox (Mac App Store)Unsandboxed Electron
The AppNative Swift with Liquid Glass, macOS 26.4 or laterMac CatalystElectron

How to Run Anima on iPhone, iPad, and Mac

  1. Pre-order Private Diffusion on the App Store; it downloads itself on release day.
  2. Open Downloads in the app and tap Anima. It is a single 2.51 GB download with nothing to configure.
  3. Pick Anima in Studio, describe the picture, and generate. It runs on your own device, so there is no queue.
Pre-order on the App Store

Questions About Anima

  • Yes. Every iPhone from the 15 Pro onward has at least 8 GB of memory, which is what Anima needs, so it runs on every supported iPhone.

  • 2.51 GB. It downloads once inside the app from Private Diffusion's own server and picks up where it left off after an interruption. On iPhone and iPad the download keeps going with the app closed; on a Mac, quitting the app stops it, and it resumes at the next launch. You can delete it to reclaim the space and fetch it again later.

  • Write tags, not sentences. Use lowercase, comma-separated tags with spaces instead of underscores, and start with who is in the picture: "1girl, solo, cherry blossoms, soft light". If you name a character, add their series and a few words about how they look. Sentences work too, but write at least two, because a very short prose prompt can give unexpected results.

    Build a prompt for Anima part by part

  • CircleStone Labs Commercial License (NVIDIA Open Model License flow-through): Your images are yours, including for commercial use, except that you may not use them to train a model that competes with Anima. Built on NVIDIA Cosmos, so the NVIDIA Open Model License applies as well; the link is on this page.

  • Yes. After the one-time download, Anima works without a connection. It turns your prompt into a finished image on the device, and you never sign in.

  • It is faster, the pictures hold up better, and nothing leaves your device. Private Diffusion ships Anima quantization aware distilled (QAD), trained against the full-size original, so it stays near-lossless at a fraction of the size. It runs on Apple's own frameworks: the GPU through MLX, including the Neural Accelerators in M5, M6, A19 Pro, and A20 Pro chips, and the Neural Engine through Core ML. ComfyUI runs models through PyTorch's MPS backend, the slowest inference framework on Apple platforms. ComfyUI's desktop app sends usage telemetry to PostHog and error reports to Datadog, with telemetry checked by default on first launch; Private Diffusion collects no data. And Private Diffusion is a sandboxed native Swift app with the hardened runtime on, where ComfyUI's desktop app is an Electron app that is not sandboxed.

  • Private Diffusion ships Anima quantization aware distilled (QAD), trained against the full-size original, so it stays near-lossless; Draw Things shrinks its models after training. It is faster, using the GPU and the Neural Engine together, plus the Neural Accelerators in M5, M6, A19 Pro, and A20 Pro chips. Private Diffusion collects no data, while Draw Things' App Store privacy label lists Purchases, Identifiers, and User Content. And it is a native Swift app built to Apple's Human Interface Guidelines, with Liquid Glass on macOS 26.4 or later, not a Mac Catalyst port.