Choosing a model · LeSysBot Docs

The model decides which tool to use when you ask in words. A bigger model gets it right more often. Pick the largest one that fits your GPU.

Using OpenAI or another remote service? Skip this page and see Settings.

Pick by GPU memory#

GPU memory	Model	Download it
4 GB	Qwen3.5 4B (the default)	`ollama pull qwen3.5:4b`
8 GB	Qwen3.5 9B ⭐	`ollama pull qwen3.5`
12 GB	Gemma4 12B	`ollama pull gemma4:12b`
16 GB	Qwen3 14B	`ollama pull qwen3:14b`
24 GB	Qwen3.5 27B	`ollama pull qwen3.5:27b`
48 GB+	Llama 3.3 70B	`ollama pull llama3.3:70b`

⭐ The best balance of quality and speed for most machines.

Switch to it#

bash

ollama pull qwen3.5
lesysbot chat --model qwen3.5     # try it for one session

Happy with it? Run lesysbot setup and pick it from the list, or set llm.model in ~/.lesysbot/config.yaml and restart the service.

Tips#

Wrong tool, or no tool at all? Try a bigger model before anything else. Qwen3.5 and Gemma4 are the most reliable at calling tools.
Out of memory? Go one size down.
First reply is slow? The model is loading. Later replies are faster.

Ollama basics#

bash

ollama list          # downloaded models
ollama ps            # models loaded right now
ollama rm NAME       # delete a model
curl localhost:11434 # prints "Ollama is running" if the server is up

Not running? sudo systemctl start ollama, or ollama serve in a terminal. More in the Ollama docs.