Anthropic's Claude Opus 5 is now available on GitLab Duo Agent Platform, targeting high-complexity engineering tasks. In GitLab's internal benchmarks, Opus 5 achieved a 93.3% task resolution rate, a 20.3-point improvement over Opus 4.8's 73.0%. It completed 100% of attempted tasks and ran 2.2% faster than Opus 4.8 at the 95th percentile. Key capabilities include multi-file refactors, coordinated multi-agent workflows with writer-verifier patterns, and precise code review with low false positives. GitLab Credits usage caps allow cost control for parallel agent workloads. Opus 5 is positioned alongside Sonnet-class models, with guidance to use Opus 5 for the hardest, highest-stakes tasks and Sonnet for routine daily work.
Table of contents
Reasoning that holds up under complexitySpeed keeps pace with depthChoose the right model for the task at handPut Opus 5 to work226 Impressions