GitLab has expanded GitLab Duo Self-Hosted to support models deployed through Microsoft Foundry. This allows organizations to run GitLab's AI development capabilities against models hosted within their own chosen Azure environment. The integration supports model families including OpenAI GPT, Anthropic Claude, Meta Llama, and Mistral. Enterprises receive more choice over model provider, deployment location, and data path.
This update is particularly relevant for organizations with data residency, sovereignty, regulatory, or network isolation requirements. Rather than sending AI requests to GitLab-managed model infrastructure, GitLab Duo Self-Hosted can use an organization's own AI Gateway and model deployments. Administrators retain greater control over where requests and responses are processed and how underlying models are deployed.
The architecture consists of three main components. These include a self-managed GitLab instance, a self-hosted GitLab AI Gateway, and one or more model endpoints hosted through Microsoft Foundry. The gateway acts as the intermediary between GitLab Duo and the selected models. This design avoids tying individual Duo features directly to a particular model provider.
Feature-level model selection is an important aspect of the integration. Organizations can use different models for different GitLab Duo capabilities. For example, teams can use a code-focused model for Code Suggestions, another model for agentic workloads, and a smaller model for higher-volume tasks. Model deployments can also be changed without fundamentally altering the GitLab development workflow.
This approach highlights a key trade-off with self-hosted AI. Giving organizations control over models and infrastructure provides greater flexibility. However, it shifts more responsibility onto engineering and platform teams. They must manage model deployments, capacity, networking, credentials, availability, and model lifecycle alongside the GitLab environment.
Model availability does not automatically equal GitLab Duo compatibility. Microsoft Foundry's catalogue can change faster than GitLab's supported-model matrix. Organizations need to verify compatibility across both platforms before selecting a model.
GitLab's approach follows a broader industry movement away from treating AI development tools and foundation models as a single bundled service. Microsoft Foundry provides access to models from multiple vendors. GitLab provides the development and DevSecOps layer around them.
Similarities exist with other enterprise development platforms. GitHub Copilot increasingly supports multiple underlying models, but its standard experience remains closely integrated with GitHub's managed service. GitLab's self-hosted model approach places greater emphasis on controlling AI infrastructure and network paths. Meanwhile, platforms such as Amazon Bedrock and Microsoft Foundry provide multi-model infrastructure but do not substitute for an integrated DevSecOps platform.
The development environment increasingly resembles a model-agnostic control layer. GitLab manages the developer workflow and AI features, while the organization determines which models sit underneath them. The significance of the announcement extends beyond another model integration.
As AI embeds deeper into software engineering, enterprises must make decisions about more than just developer AI capabilities. They must consider where models run, where source code and prompts travel, who controls credentials, and which jurisdictions process the data. GitLab's Microsoft Foundry integration addresses part of that problem by allowing the AI model layer to sit within an organization's chosen Azure environment.
Organizations should verify compatibility across both platforms before selecting a model.
