Go 1.25 Introduces Experimental Green Tea Garbage Collector
The Go programming language team has announced the inclusion of a new experimental garbage collector, named Green Tea, in the upcoming Go 1.25 release. This feature can be enabled by setting the GOEXPERIMENT=greenteagc flag during build time. According to the developers, many workloads experience approximately a 10% reduction in time spent on garbage collection, with some seeing improvements of up to 40%. Although labeled experimental, the Green Tea collector is described as production-ready and is already deployed within Google's infrastructure. The team encourages developers to test the new collector and provide feedback, noting that while most benefit, some workloads may not see significant gains. Based on current data, the developers plan to make Green Tea the default garbage collector in Go 1.26. Users are invited to report issues or successes through specific channels on the Go project's issue tracker. This announcement stems from a presentation given at GopherCon 2025, highlighting advancements in Go's memory management efficiency.
Wire timeline
Go 1.25 Introduces Experimental Green Tea Garbage Collector
The Go programming language team has announced the inclusion of a new experimental garbage collector, named Green Tea, in the upcoming Go 1.25 release. This feature can be enabled by setting the GOEXPERIMENT=greenteagc flag during build time. According to the developers, many workloads experience approximately a 10% reduction in time spent on garbage collection, with some seeing improvements of up to 40%. Although labeled experimental, the Green Tea collector is described as production-ready and is already deployed within Google's infrastructure. The team encourages developers to test the new collector and provide feedback, noting that while most benefit, some workloads may not see significant gains. Based on current data, the developers plan to make Green Tea the default garbage collector in Go 1.26. Users are invited to report issues or successes through specific channels on the Go project's issue tracker. This announcement stems from a presentation given at GopherCon 2025, highlighting advancements in Go's memory management efficiency.
The Go Blog