Skip to content

Model Setup

After installing Confidential Translator, you need at least one LLM model for translations.

Model Size Best For Quality
qwen3:14b-q4_K_M ~8 GB Best balance of quality & speed ⭐⭐⭐⭐⭐
qwen3:8b ~5 GB Fast translations, good quality ⭐⭐⭐⭐
qwen2.5:7b ~4 GB Lightweight, decent quality ⭐⭐⭐

Pull a Model

Via the UI

  1. Navigate to the Models page in the sidebar
  2. Enter the model name (e.g. qwen3:14b-q4_K_M) in the pull field
  3. Click Pull — progress is shown in real-time

Via CLI

ollama pull qwen3:14b-q4_K_M

Setting the Default Model

  1. Go to Settings → Translation Settings
  2. Set the Default Model field to your preferred model name
  3. This model will be used automatically for all translations

Custom Model Path

If you store models on an external drive, update the Model Storage Path in Settings → Ollama Engine. Set the OLLAMA_MODELS environment variable accordingly.