Fangyuan025/Chaty v2.3.8

on GitHub

FLUX.2 [klein]

FLUX.2 [klein] 4B and 9B, distilled and base, now load and draw. A klein GGUF
is the denoiser alone; Chaty recognises it, offers the Qwen3 encoder and the
FLUX.2 VAE it needs in the same download, and draws at the settings it was
made for — four steps at CFG 1 for the distilled ones. klein 4B draws
1024×1024 in about a minute on an M4 Pro.

Errors in the language you picked

Every message from the backend is now in the language the UI is in, from the
first one at startup on, instead of Chinese until a coding turn started, or
both languages at once. An engine's own words come after a line in yours, and
an image model missing a file says which one.

Linux, and the store

Thanks to @davidecelentano: on Linux Chaty picks one GPU through Vulkan —
discrete over integrated — and sizes offload by what the driver says is free;
music plays in the AppImage. The store gains a Text generation filter, links
to Hugging Face, and a fit hint that knows whether a model fits the GPU, or,
on a Mac, fits in memory with everything it downloads.

Install

macOS first launch — "Apple could not verify…" / "damaged"

Chaty is ad-hoc signed but not notarized (no paid Apple Developer
account), so Gatekeeper flags it on first launch. The app is safe —
everything runs locally. Clear the download quarantine once, in Terminal:

xattr -dr com.apple.quarantine /Applications/Chaty.app

Then open Chaty normally. (Alternatively: try to open it once, then go to
System Settings → Privacy & Security → Open Anyway.)

What's Changed

• Improve Linux Vulkan support, model catalog and media playback by @davidecelentano in #24

New Contributors

• @davidecelentano made their first contribution in #24

Full Changelog: v2.3.7...v2.3.8