GitLab’s Transcend virtual event on June 10–11, 2026, served as a major platform update for the company, showcasing a pivot toward “agentic engineering at enterprise scale.” The headline announcement is an expanded collaboration with Google Cloud to deliver a fully managed DevSecOps platform, but the event also featured significant updates to GitLab’s core code management engine, an AI-powered context graph, and a new licensing model [5, 22].
A Fully Managed GitLab on Google Cloud
GitLab and Google Cloud are now working with a network of certified managed service providers (MSPs), including Beyond and Digital Future, to offer a fully managed GitLab DevSecOps platform on Google Cloud [22, 24, 33]. This move targets enterprises with strict data sovereignty and residency requirements, allowing them to keep code, pipelines, and security data within mandated geographies while outsourcing platform operations [24, 40].
Organizations can purchase GitLab and the Duo Agent Platform directly through the Google Cloud Marketplace and apply that spend toward their existing Google Cloud commitments ![]()
. This builds on an April 2026 agreement that allowed GitLab Duo usage to count against Google Cloud spending [22, 38].
Google’s Latest AI Models Come to GitLab Duo
Availability of Google’s newest models within GitLab was a major focus. The latest versions of Gemini, including Gemini 3.5, are now available in the GitLab Duo Agent Platform [17, 22, 33]. For self-hosted customers, Google’s open model family, Gemma 4, is also available [22, 33]. This reinforces GitLab’s broader support for a multi-model strategy: GitLab Duo documentation confirms that users can select from models provided by Anthropic, Meta, Mistral, and OpenAI through various cloud providers [19, 18].
Google’s own documentation supports the general availability of Gemini 3.5 Flash ![]()
and the release of Gemma 4 models like the 26B A4B IT variant ![]()
.
Next-Generation Source Code Management
GitLab announced a “next-generation source code management” engine, now in private beta [5, 6]. The new Git engine is rebuilt for agent-scale concurrency, replacing traditional repository clones with structured API access to project data. GitLab claims this enables AI coding agents to complete tasks up to 50 times faster per agent compared to previous methods [4, 6].
GitLab Orbit: A Unified Context Graph
Perhaps the most ambitious AI infrastructure piece is GitLab Orbit, a context graph that spans the full software lifecycle [5, 6]. Now in public beta, Orbit is designed to give AI agents a unified understanding of code, issues, pipelines, and security data. Internally, GitLab reports that Orbit delivers 11x faster agent responses while consuming 4.5x fewer tokens [4, 6]. This is intended to dramatically reduce AI hallucinations and irrelevant outputs by keeping agents grounded in the complete project context.
Agents for Security and Governance
To balance the speed of AI agents with enterprise control, GitLab also introduced a private beta for Agents for Security and Governance [5, 6]. This layer provides identity, policy, audit, and approval controls around every agent action, helping organizations meet compliance standards even as they adopt autonomous coding and operations ![]()
.
GitLab Flex: A New Licensing Model
GitLab also announced GitLab Flex, a purchasing model now accepting orders [5, 53, 56]. Instead of separate contracts for seats and add-on usage, GitLab Flex uses a single annual dollar commitment that covers platform seats, GitLab Credits, and eligible usage-based capabilities [49, 53]. Organizations can adjust their allocation of seats and credits month-to-month without requiring new contracts or amendments [49, 56].
GitLab Credits, priced at roughly $1 per credit on-demand, serve as the internal currency for accessing the Duo Agent Platform and other usage-based features [48, 61]. The consumption-based nature of this model was underscored during GitLab’s Q1 2027 earnings call, where the company noted the Duo Agent Platform had reached nearly $20 million in paid consumption run rate ![]()
.
What This Means for the Competitive Landscape
While the event clearly positions GitLab and Google Cloud as deepening their alliance, available sources do not support direct competitive superiority claims over Microsoft’s GitHub/Azure stack or AWS CodeCatalyst. GitLab’s documented advantages include flexible licensing, multi-model AI support, and self-hosted deployment options for sensitive environments [4, 19, 53]. A managed, partner-delivered model on Google Cloud adds another option for enterprises, but no public evidence was provided to benchmark this model against the integrated stacks of Microsoft or Amazon.
