Google's Gemini 4 Argon is now accessible on GitLab's Duo Agent Platform, designed to carry long, multi-step engineering tasks through to review.
The model aims to enable faster work across the software lifecycle while keeping risk and cost within enterprise controls and guardrails.
Argon lets teams delegate extended actions and receive outputs that can be reviewed as part of the development process.
DESIGNED FOR LONG-HORIZON TASKS WITH ENTERPRISE GUARDRAILS
Argon is trained to respect system boundaries, to pause for confirmation before irreversible steps, and to indicate when something cannot be found or completed. TechStaged has also covered GPT-6 Astra on GitLab Duo Agent Platform delivers faster runs and fewer tokens.
This enables teams to hand off complex tasks while maintaining control over critical decisions.
BENCHMARKS AND RELIABILITY FOR LONG-HORIZON WORK
In GitLab's internal evaluation, Gemini 4 Argon completed 93.3% of GitLab’s software engineering benchmark tasks.
Argon sustains its reasoning across many steps—reading files, running commands, and editing code—to completion.
Additional findings show 73.3% of benchmark tasks were resolved, with 93.8% of runs that hit a tool error or timeout still finishing.
- 73.3% of benchmark tasks resolved
- 93.8% of runs with tool error/timeout completed
- Outputs arrive in a merge request for reviewer context
MODEL OPTIONS AND PAIRING WITH FLASH MODELS
Gemini 4 Argon is described as the most performant Gemini model on GitLab Duo Agent Platform, optimized for long-horizon actions and loosely defined tasks.
It can be paired with Flash models for lower-cost, high-volume workloads so teams can match workloads to the right Gemini tier.
AVAILABILITY AND PRICING
Argon is available now on GitLab Duo Agent Platform.
Pricing includes a promotional rate of 2.86 calls per credit through November 30, 2026; starting December 1, 2026, standard pricing applies at 1.33 calls per credit.
GitLab Premium and Ultimate customers can enable Duo Agent Platform and try Argon using GitLab Credits included with their subscription; new users can start a free trial.
WHAT THIS MEANS FOR TEAMS
Agent results land in a merge request with full context for review, allowing engineers to focus on verification and decision-making rather than monitoring long-running runs.
Argon runs inside the same context layer, policy checks, and audit trail that cover every agentic action across the software lifecycle.
WHAT HAPPENS NEXT
Argon is available now; teams can select Gemini 4 Argon in model selection within GitLab Duo Agent Platform.
RELATED COVERAGE
- GPT-6 Astra on GitLab Duo Agent Platform delivers faster runs and fewer tokens
- GitLab arms developers with cost controls for AI agents via Duo Agent Platform and Orbit
- GitLab Credits Bring Usage-Based Pricing to Duo Agent Platform
- GitLab’s agent orchestration cuts the wait between development stages with Duo Agent Platform
- Developer Tools articles
SOURCES
- GitLab: Gemini 4 Argon on GitLab: Built for long-running agentic work Published · Primary source
