Local AI app Ollama now supports image generation, starting with FLUX.2 [klein] and Z-Image-Turbo
On January 20, 2026, it was announced that Ollama , an app that can run various AI locally, now supports image generation AI as an experimental feature. At the time of writing, the macOS version of Ollama is capable of generating images using
Image generation (experimental) · Ollama Blog
https://ollama.com/blog/image-generation
Ollama is an app developed as a tool for easily running large-scale language models in a local environment. In July 2025, a GUI app was released that allowed users to download and run various open models, such as Llama, Gemma, and gpt-oss, using an easy-to-understand UI. As of July 2025, it supported image loading, allowing users to load images into the multimodal AI and ask for details.
A GUI app version of the popular local AI tool 'Ollama' has been released, allowing you to chat by running many large-scale language models locally - GIGAZINE
On January 20, 2026, Ollama also supported image generation AI as an experimental feature. At the time of writing, the macOS version of Ollama supports image generation using FLUX.2 [klein] and Z-Image-Turbo.
Below is an example of an image generated by Z-Image-Turbo. If you use a terminal emulator that supports image display, such as Ghostty or iTerm2, you can display the generated results inline.
FLUX.2 [klein] is available with 4 billion parameter models and 9 billion parameter models, and can also run fp8, fp8, and bf16 quantized models.
Tags · x/flux2-klein
https://ollama.com/x/flux2-klein/tags
Z-Image-Turbo supports running fp8 and bf16 models.
Tags · x/z-image-turbo
https://ollama.com/x/z-image-turbo/tags
Image generation on Windows and Linux will be supported at a later date, and image editing functions will also be supported in the future.
Related Posts:
