Configuration

All LlamaTray settings with their defaults: GPU layers, context size, port, sampler presets, mmproj, router options and profiles.

Settings overview

Setting Description Default
Run mode Single Model or Router Mode Single Model
GPU Layers Number of layers to offload to the GPU 99
Context Size Context window size (512 – 1,000,000) 32768
Port Server port (1024 – 65535), bound to 127.0.0.1 8080
Sampler Preset Neutral, balanced, creative, precise or custom Custom
Extra Parameters Any additional llama-server flags (optional)
mmproj File Multimodal projector file (Single Model mode only) (optional)

Router mode settings

Setting Description Default
Models Directory Directory containing the GGUF models exposed by the router (required)
Auto-load Models Automatically load router models when required Disabled
Jinja Enable Jinja chat templates Enabled

Sampler presets

Presets map directly to llama-server sampling flags:

Preset Flags
Neutral --temp 0.7 --top-p 0.9 --top-k 40 --min-p 0.0 --repeat-penalty 1.0
Balanced --temp 0.5 --top-p 0.95 --top-k 20 --min-p 0.0 --repeat-penalty 1.05
Creative --temp 1.0 --top-p 0.99 --top-k 100 --min-p 0.0 --repeat-penalty 1.0
Precise --temp 0.1 --top-p 0.5 --top-k 10 --min-p 0.0 --repeat-penalty 1.1

Editing the extra parameters manually switches the preset back to Custom.

Profiles

Profiles are named configurations covering both Single Model and Router setups:

  • Save the current form values as a new profile, or update an existing one.
  • Load applies a profile to the form; delete removes it.
  • Profiles are stored in ~/.llamatray/profiles.json.

Application preferences (mode, language, minimize-to-tray, …) are stored separately in ~/.llamatray/config.json. Both are restored automatically on startup and are never removed by the uninstaller.

Language

The interface is available in Turkish and English; switching language applies immediately and is persisted with your settings.

llama-server discovery

LlamaTray locates llama-server in this order:

  1. llama-server on your PATH
  2. /usr/bin/llama-server, /usr/local/bin/llama-server
  3. ~/.local/bin/llama-server
  4. ~/llama.cpp/build/bin/llama-server
  5. ~/llama.cpp/server/llama-server

If none is found, use the built-in llama.cpp manager or install llama.cpp yourself — no configuration inside LlamaTray is required.

Server binding

The server always binds to 127.0.0.1 (localhost) on the selected port. Before starting, LlamaTray checks that the port is free and cleans up stale llama-server processes occupying it.