Msty Studio 3.x · Desktop guide

Import Models

Use a local GGUF file or a compatible safetensors model directory.

Import files you already have

  1. Open Settings > Local Models.
  2. Choose Import.
  3. Select a GGUF file, or choose the directory option for a safetensors model.
  4. Follow any required engine setup and wait for import to finish.
  5. Find the model under Installed and select it in a conversation.

Keep all required configuration, tokenizer, and weight files together when importing a safetensors directory. A single weight file may not contain everything the engine needs.

Which engine receives the model?

  • GGUF imports use an installed LLaMA.cpp engine when available; otherwise they use Ollama, with setup offered if needed.
  • On Apple Silicon, an existing MLX provider can import a compatible MLX model directory.
  • Other supported safetensors imports use Ollama. Compatibility depends on the model format and architecture.

Importing a file does not make an unsupported model architecture compatible. If import fails, check the reported error and confirm that the complete model is supported by the selected engine.

To download a model directly instead, use Local Models.

Local Models
Discover, install, and manage models running on your computer.
External Model Servers
Connect models served by Ollama, LM Studio, or another compatible endpoint.