Overview

Artemis Desktop brings systematic code optimisation to your laptop, desktop or workstation. Connect an open source model running on your own machine and use Artemis to explore, test and measure alternative implementations. Model inference runs locally, so model calls do not incur external API token costs.

Artemis is an optimisation platform that improves the performance of models, agents and code without sacrificing accuracy. It systematically explores candidate solutions, executes them and measures their results to identify improvements that are supported by evidence.

What is Artemis Desktop?

Artemis Desktop connects to a locally running large language model (LLM) through an API endpoint. It provides access to the same workflows available in the full Artemis SaaS platform. Artemis agents run on TurinTech infrastructure, and model inference runs on your own hardware, so Artemis Desktop needs an internet connection.

The recommended model is Qwen3 Coder 30B A3B Instruct. See Choose a model for the other recommended model and serving engines. You are responsible for downloading, preparing and running your chosen model. You can set it up before or after installing Artemis Desktop.

Key benefits

• No external model token costs: Run model inference locally without paying per token to an external model provider.
• Run on your own hardware: Use your existing laptop, desktop or workstation to run local models and perform optimisation work.
• Evidence-based optimisation: Explore alternative implementations, execute candidates and measure their performance rather than relying on AI-generated suggestions alone.
• Flexible model setup: Connect to compatible open source models through a locally running API endpoint.
• Your choice of deployment: Use Artemis Desktop alongside other deployment options, including Artemis SaaS, private cloud and on-premise installations.

Hardware considerations

Local models are limited by the memory and compute resources available on your machine. Larger or more capable models may require more powerful hardware. Artemis SaaS provides access to frontier models for workloads that benefit from them.

How it works

1. Start a model server on your machine, such as OpenVINO or llama.cpp serving Qwen3 Coder 30B A3B Instruct. See Run with OpenVINO or Run with llama.cpp.
2. Add it as a source under Models in Artemis Desktop, using the server's API endpoint. See Connect a model.
3. Run a workflow. Artemis agents run on TurinTech infrastructure. Whenever an agent needs a model, the request goes through Artemis Desktop to your model server, so inference runs on your hardware.

Get started

• Install Artemis Desktop: review the prerequisites and installation steps.
• Set up your model: choose a model, run the model server and connect it to Artemis Desktop.
• Quick start: run your first Discovery on a Python project, in the app or with a coding agent.
• Workflows: learn about Maintain, Discover, Plan, Branches and AI Agents in Artemis Desktop.
• Troubleshooting: fix connection and model server problems.

For other ways to run Artemis, see Deployment options.