On first run vtmate creates ~/.vtmate/settings and ~/.vtmate/agents
with working defaults, pointed at ollama running llama3.2:3b -
so the only thing left for you to install is ollama itself and that one model.
1. Install ollama
curl -fsSL https://ollama.com/install.sh | shOn macOS or Windows, use the installer from ollama.com/download instead.
2. Pull the default model
ollama pull llama3.2:3b3. Run vtmate
vtmateThat's it: ollama listens on http://127.0.0.1:11434, which is exactly what the default agents already point at.
~/.vtmate/settings as created[general]
selected_agent = main agent
[daemon]
llm_background_ptt_combo = ctrl+alt+a
tts_background_combo = ctrl+alt+r
stt_and_paste_background_ptt_combo = ctrl+alt+s
llm_background_reset = ctrl+q~/.vtmate/agents as created
vtmate ships with 7 ready-to-use personas so you have something to switch between right
away with ARROW_LEFT / ARROW_RIGHT:
main agent (the default), explainer, planner,
Ptahhotep, Aristoteles, Budda and
Jesus Christ - each with its own voice and system prompt, all on the same
ollama / llama3.2:3b model. The first one, in full:
[agent]
name = main agent
language = en
tts = supertonic3
voice = M1
voice_speed = 1.1
provider = ollama
baseurl = http://127.0.0.1:11434
model = llama3.2:3b
system_prompt = @concise_assistant
sound_threshold_peak = 0.12
end_silence_ms = 2500
ptt = true
whisper_model_path = ~/.whisper-models/ggml-tiny.bin
Its system_prompt = @concise_assistant pulls in a shared
[system_prompt] block also written to the same file - see
Reusable system prompts on the Configure Agents page.
Every field here is explained in full in
Agent fields & providers.
Swap ollama pull llama3.2:3b for any other ollama model and update
model in each agent (or in the Control+S popup)
to match. To use llama-server, a hosted API, or a CLI subscription instead of ollama, see
Agent fields & providers for the
full provider / baseurl / api_key reference.