Offline local model staging

Cute LM makes local AI feel calm, private, and easy to launch.

Cute LM is a local language model launcher for people who want compatible models running from their own machine without cloud inference, mystery endpoints, or built-in download systems.

Download Cute LM Support on Ko-fi

Cute LM is designed to stay offline-first. Want to help keep it free? Support development at ko-fi.com/wordblocklabs.

Cute LM is built for people who already have their models and want a clean local launcher, not another account system, marketplace, or cloud dependency.

What It Is

  • A privacy-first local model launcher for offline use.
  • A GUI for loading model folders and LoRA folders from your own drive.
  • A localhost runtime shaped for apps that expect familiar LM Studio-style defaults.
  • A Mac-first release centered on the tested Qwen MLX lane.

What It Is Not

  • Not a cloud AI service.
  • Not a built-in model downloader.
  • Not a browsing tool for models or adapters.
  • Not yet a fully finished every-backend, every-platform runtime.

Current release

What Cute LM can do right now

Run local MLX models

Load a compatible MLX model from disk and launch it directly from your machine.

Use one or two LoRAs

Load Slot A, Slot B, or both together when the selected pair is compatible with Cute LM’s current composition path.

Stay on a simple endpoint

Keep the familiar local endpoint at 127.0.0.1:1234 for compatible apps.

Limitations

What it cannot do yet

  • You must download models and LoRAs yourself before using Cute LM.
  • GGUF is recognized in the interface, but full GGUF runtime support is still in progress.
  • The current production runtime path is Mac-first.
  • Dual-LoRA launch depends on both adapters being compatible with the current composition rules.

Best practices

How to get the cleanest experience

  • Keep base models in one easy-to-find folder.
  • Keep LoRAs in a separate easy-to-find folder.
  • Load those folders manually from Cute LM instead of relying on temporary paths.
  • If Cute LM reports a missing file, check that the selected folder contains the full model or adapter files.

Support and updates

Solo-developed, steadily improved, and easy to report issues on.

Cute LM is a solo-developed project. Updates are released as issues are discovered, fixes are completed, and new features are added.

For support, bug reports, and project updates, visit WordBlockLabs.com. Bug reports are most useful when they include a screenshot and the exact steps you took.

Looking for other WordBlock Labs products? Visit CuteLM.wordblocklabs.com for Cute LM and Solo.wordblocklabs.com for SoloRoleplayer//.

Get Cute LM Support Development