chaiNNer at a glance
Checked 16 Oct 2026chaiNNer chains image operations as nodes: load a folder or video, upscale with a model, then resize, sharpen or save. It loads community models (e.g. from OpenModelDB) through its Spandrel library, which also covers GFPGAN and CodeFormer face restoration. It ships no default model, so the weights licence is whatever you load. Code is GPL-3.0. It supports Nvidia, AMD, Intel and Apple GPUs, with CPU fallback. It is not a one-click tool; you build the chain yourself. Last release v0.25.1 (Oct 2025); commits continue.
RepoQuarry. “chaiNNer.” Checked 16 Oct 2026. https://repoquarry.dev/projects/chainnerHow much of Topaz Gigapixel does it replace?
4 of 6 jobs. Covers these jobs with models you choose and chains you build; no generative upscaling models or Photoshop plugin.
For recover and redefine low-quality or compressed images, Photoshop and Lightroom plugin, see other Topaz Gigapixel alternatives. You may also need to build an interface, an API server and hosting yourself.
Facts, with sources
Every answer links to where we found it. We read the documentation; we did not install it.
You don't even need to have Python installed, as chaiNNer will download an isolated integrated Python build on startup.github.com · chaiNNer#installation ↗ today
Architectures with restrictive licenses (e.g. non-commercial) are implemented in the `spandrel_extra_arches` package.github.com · spandrel#readme ↗ today
The `spandrel` package only contains architectures with permissive and public domain licenses (MIT, Apache 2.0, public domain), so it is fit for every use case.github.com · spandrel#readme ↗ today
CPU: All frameworks support CPU-only mode as a fallback.github.com · chaiNNer#gpu-support ↗ today
ChaiNNer is also cross-platform, meaning you can run it on Windows, MacOS, and Linux.github.com · chaiNNer#readme ↗ today
Download the latest release from the Github releases page and run the installer best suited for your system. Simple as that.github.com · chaiNNer#installation ↗ today