GitLab Duo Self-Hosted Connects to Microsoft Foundry

GitLab Duo Self-Hosted allows organizations to connect AI coding tools to models running on infrastructure they choose, giving them control over hosting, region, network path, and credentials.

The solution is particularly useful for teams under data sovereignty or regulatory constraints who need to know which network handles their source code before adopting an AI tool.

GitLab Duo Self-Hosted can be connected to models hosted in Microsoft Foundry, a general-purpose serving platform that offers a range of large language model (LLM) families, including OpenAI GPT, Anthropic Claude, and Meta Llama.

The setup process involves deploying one or more model endpoints through Microsoft Foundry, configuring the AI Gateway service to route requests between GitLab and the chosen LLM backend, and assigning models to individual features in GitLab Duo.