Choosing a Model

Choosing a Model

Draggy runs three models, and they do different jobs. Most of the time you only think about the first one.

The chat model

This is the one that answers you in the main window and calls tools. It is the big one, and it is what the first-launch download fetches.

Draggy picks it from your VRAM:

VRAM	Model
under 1 GB	SmolLM2 360M
1 GB	Qwen 3 0.6B
1.5 GB	Qwen 3 1.7B
3 GB	Llama 3.2 3B
4 GB	Phi-4 Mini
5 GB	Qwen 3 4B
8 GB	Qwen 3 8B
12 GB	Phi-4
14 GB	Qwen 3 14B
20 GB	Qwen 3 32B
48 GB and up	Qwen 3 235B MoE

These are starting points, not rules. Change it whenever you like in Settings → Models, or from the model name at the top of the chat window. The settings panel shows how much of each model will fit in VRAM and how much would spill onto the processor, which is the number that decides whether a model feels usable.

A model that reports "100% GPU" will answer at reading speed. One at 60% will still work, several times slower. Below that it is usually better to step down a size.

Anything with -cloud in its name is hidden. Draggy runs locally by definition.

The talk model

Voice mode uses its own model, and by default not the same one. A spoken reply is two sentences long and has to start arriving within a few hundred milliseconds of you falling silent, which is a completely different problem from writing a good paragraph. The chat model is loaded for context length and reasoning depth, and you pay for both on every turn.

So voice mode picks a smaller one:

VRAM	Model
under 2 GB	SmolLM2 360M
2 GB	Llama 3.2 1B
4 GB	Qwen 3 1.7B
6 GB	Llama 3.2 3B
8 GB	Qwen 3 4B
16 GB and up	Qwen 3 8B

The ladder deliberately stops at 8B. Past about four billion parameters a spoken answer is not noticeably better, and the extra weights are paid for on every single turn.

The picker in voice mode has Automatic at the top, which is this table, and below it every model you have installed. If you pick one explicitly it is remembered. If you later delete that model, Draggy quietly goes back to automatic rather than downloading it again behind your back.

When automatic picks something you do not have, it downloads it the first time you press Start talking, with a progress bar and a cancel button. The size is shown under the button before you commit to it.

The helper model

A very small model — 360M to 1.7B — that reads your question and decides whether answering it needs the web, your own documents, or nothing. That decision is cheap enough to make before the real model starts, and it saves the chat model from reasoning about its own tools.

You will rarely think about this one. It is downloaded automatically, weighs a few hundred megabytes, and can be turned off entirely in Settings → Tools if you would rather the main model decided for itself.

Running several at once

Ollama keeps models in memory for a while after use. Chat, voice and the helper are three different models, and on a card with enough room all three sit resident together, which is why replies stay fast when you switch between them.

On a smaller card Ollama will evict one to make room for another, and you will feel that as a pause on the first reply after switching. Picking a smaller talk model is usually the fix.