Grafana Mimir 3.0 introduces significant enhancements to this open-source, horizontally scalable time series database, primarily focusing on improved reliability, performance, and cost efficiency. The new version features a decoupled architecture that separates read and write operations, employing Apache Kafka as an asynchronous buffer to enable independent scaling paths, which results in faster queries and smoother ingestion processes. The Mimir Query Engine, a key component of the update, offers a more memory-efficient approach by streaming query results rather than loading entire datasets, reducing peak memory usage by up to 92% and ensuring compatibility with Prometheus' PromQL. These architectural advancements, along with reduced resource consumption by up to 15%, make Mimir 3.0 a robust solution for managing large-scale data metrics, and it is available both as an open-source project and through the Grafana Cloud Metrics service. The release is a collaborative effort, driven by community feedback and contributions, showcasing Grafana Labs' commitment to advancing open observability solutions.