GitLab launched a managed DevSecOps platform on Google Cloud with Gemini 3.5 Flash

GitLab announced the launch of a fully managed version of its platform on Google Cloud with integrated Gemini 3.5 Flash and Gemma 4 models. Teams no longer need to choose between data sovereignty and access to cutting-edge AI—now both are available in a single managed platform.

Managed GitLab on Google Cloud

GitLab is available as a fully managed service on Google Cloud through certified partner providers (MSPs)—companies Beyond and Digital Future. Organizations transition to a scalable DevSecOps architecture without losing control over where code, pipelines, and security data are stored. This builds on the April 2026 partnership: at that time, GitLab enabled calling Google models through the Duo Agent Platform and counted their usage against existing Google Cloud commitments.

Now fully managed hosting and an accelerated model roadmap have been added. The partner provider assumes operational load and SLA compliance—the customer team does not handle infrastructure support. This is particularly important for organizations with data localization requirements: fintech, government sector, healthcare.

Data sovereignty and access to the latest AI models no longer conflict.

Gemini and Gemma: Model for the Task

GitLab's Duo Agent Platform now runs two model families:

• Gemini 3.5 Flash — fast tasks with low latency: code autocomplete, instant reviews, test generation
• Gemini (senior versions) — deep reasoning for architectural decisions, refactoring large codebases, complex security analysis at the repository level
• Gemma 4 — open-weight model for self-hosted environments: all requests and responses remain strictly within company infrastructure

GitLab is part of Google's early access program for Gemini, so new model versions appear in Duo immediately after release—without additional procurement processes from the customer. For regulated environments, Gemma 4 integrates into the Duo Self-Hosted architecture: the team deploys its own AI Gateway, and no request leaves the perimeter—whether on-premises or private cloud. Different tasks require different models—the lineup provides clear choice without the need to assemble a solution from disparate tools.

Data Control, Audit, and Unified Budget

Three key arguments for enterprise customers:

• Data under your control. Code, pipelines, and security data are stored in your chosen infrastructure—even in fully managed mode.
• Auditability out of the box. Every agent action, merge request, and scanner finding is visible to compliance teams through GitLab's built-in tools. Manageability doesn't disappear when an agent takes on part of the workflow.
• Unified billing. GitLab and the Duo Agent Platform through Google Cloud Marketplace allow expenses to be charged against an already-established Google Cloud commitment—platform, inference, and infrastructure in one bill, without reconciliation between vendors.

"Most teams sacrifice one thing: sovereignty without modern AI or modern AI without control.

This partnership was created to provide both," — from the GitLab announcement.

On top of this, native cost control tools remain: usage dashboards, model selection policies, and GitLab Credits—predictable spending for the finance department answering: "What did we spend on AI this quarter and what did we get in return?"

What This Means

The GitLab + Google Cloud combination addresses the main contradiction in enterprise development: powerful AI models and complete data control are no longer mutually exclusive. For fintech, government sector, and healthcare with strict data residency requirements, this opens a path to agentic DevSecOps without compromise on regulatory requirements. And unified billing through Google Cloud Marketplace removes operational complexity for teams already working in the Google ecosystem.