LlamaSwift is a native macOS SwiftUI wrapper for an external llama-server executable. It discovers local GGUF models, stores launch profiles, validates options against the installed binary, and manages server processes and logs.
- Native SwiftUI interface on macOS 14+
- User-installed
llama-serverexecutable - Recursive GGUF discovery, including split GGUFs, mmproj files, and draft/MTP companions
- Typed server settings plus validated custom arguments
- Port conflict checks, health-based readiness, logs, restart, and shutdown cleanup
Safetensors conversion, model downloads, chat, a bundled llama.cpp binary, and in-process llama.cpp linking are intentionally out of scope. Safetensors-only directories are ignored because llama-server cannot run them directly.
swift build
swift testFor a normal .app build, open LlamaSwift.xcodeproj and run the LlamaSwift scheme. Configure the executable and model library from LlamaSwift → Settings.
Released under the MIT License. Copyright © 2026 Alan Beltran Pozo.