Table of Contents
Waiting 13 hours for a merge request is a massive bottleneck, but the new GitLab AI code reviews aim to fix this self-defeating loop. The widespread adoption of AI coding tools has inadvertently caused code review times to spike by 91%, leaving developers waiting on approvals while backlogs pile up. To combat this, the newly released GitLab 18.10 and 18.11 updates introduce flat-rate pricing, strict spending caps, and free-tier access to the Duo Agent Platform. This shift is designed to help development teams ship code faster without breaking the bank on unpredictable token-based AI models.
Flat-Rate Pricing and Free-Tier AI Access
The headline change in version 18.10 is the introduction of a flat $0.25 fee per automated code review, regardless of the merge request's size or complexity. According to GitLab, competing tools often charge between $15 and $25 per review using token-based models. This high cost forces teams to ration AI reviews for only the most critical changes, which directly causes the massive backlogs. At a quarter per review, developers no longer have a financial reason to skip automated checks on smaller, low-stakes commits.
Additionally, free-tier users on GitLab.com can now access the Duo Agent Platform by purchasing GitLab Credits. Instead of traditional seat-based licensing, these credits are allocated at the group level and sold in monthly blocks. Group owners are provided with a usage dashboard to monitor exactly which agents and workflows are consuming the most credits.
Actionable Steps: Managing Budgets and Security
With the shift toward a pay-as-you-go model, version 18.11 introduces strict budget guardrails to prevent unexpected expenses. Platform administrators can now enforce hard credit caps across their organizations.
- Subscription-Level Limits: Billing account managers can set a hard monthly limit for the entire organization to ensure budgets are never exceeded.
- Per-User Allocations: Administrators can establish individual credit limits to prevent a handful of power users from burning through the team's monthly pool.
- Vertex AI Integration: Customers using Google Cloud can route model calls through Vertex AI via GitLab's AI Gateway, allowing AI tool usage to fall under existing cloud agreements.
- SAST False Positive Detection: For Ultimate customers, the Duo Agent Platform now scores new critical and high-severity static analysis findings based on their likelihood of being false positives, surfacing this data directly in the Vulnerability Report.
The Shift Toward Consumption-Based AI
The transition from seat-based licensing to a credits model marks a significant pivot in how DevSecOps platforms monetize AI. On Reddit's GitLab community, users have already noted that sales representatives are phasing out Duo Pro and Duo Enterprise licenses in favor of this pay-as-you-go structure. While this raises valid concerns about how existing contracts will be honored, the move directly addresses the unpredictability of modern AI tool usage.
By dropping the cost of automated reviews to a flat $0.25 and implementing hard spending caps, GitLab is aggressively targeting the 13-hour merge request bottleneck. If teams no longer have to ration their AI usage to avoid massive token overages, we could see a dramatic reduction in deployment delays. However, the true test will be whether enterprise customers embrace this consumption model or push back against the loss of predictable, fixed-seat billing. Full details are available in the 18.10 and 18.11 release notes.
