Kokon icon
Kokon icon

Kokon

 1 like

Edit photos by describing the change, on your own GPU. A free, open-source alternative to Nano Banana and ChatGPT image editing: photos stay on your computer, no daily limit. One installer for Windows, Mac and Linux sets up the best open model (Qwen-Image 2.1). MCP server for AI agents.

Kokon screenshot 1

Cost / License

Application type

Since

  • New

Updated

  • Updated Oct 2026

Platforms

  • Windows
  • Mac
  • Linux
  • Docker
  • Self-Hosted
1like
0articles
Save

Features

Properties

Features

Highlights All activities

Kokon News & Activities

Recent activities

Kokon information

AlternativeTo Categories

Photos & Graphics, AI Tools & Services, System & Hardware

GitHub repository

  •  233 Stars
  •  33 Forks
  •  10 Open Issues
  •   Updated  
View on GitHub
Kokon was added to AlternativeTo by GabrieleLozupone on and this page was last updated .

No comments or reviews, maybe you want to be first?

Official Links

What is Kokon?

Kokon edits and creates photos with AI entirely on your own computer. Open a photo, paint an area or just describe the change ("remove the scooter", "make it a snowy night", "change the sign to say…"). Kokon picks the open model (Qwen-Image 2.1 Turbo, in Unsloth's Q4_K_M build), downloads it and sets it up for your graphics card or Mac. No Python, no node graphs, no account, no cloud, no usage limits.

Every edit is a new version with a before/after slider. Kokon also expands photos, combines several images, upscales up to 4K, opens iPhone HEIC photos and turns a short idea, in any language, into a detailed prompt (Prompt+).

Since version 0.11 you can use it from your phone, tablet or laptop in a browser on your home network, behind a password, or run it as a headless Linux server or a Docker image for NVIDIA cards. Apple Silicon Macs get an experimental engine of their own. A built-in MCP server lets AI agents such as Claude or Cursor create and edit images on your machine.

Needs a 12 GB graphics card (8 GB in beta; NVIDIA, AMD or Intel Arc) or an Apple Silicon Mac with 16 GB. Measured: about 24 s per edit on an RTX 5070 Ti; on a 16 GB M2 Pro about 4 minutes per image, or about 1 minute with the experimental Mac engine. Free and open source (AGPL-3.0). Images are for non-commercial use because of the model's license.

Kokon Videos

Show 1 more video