GrowthBook provides the most feature-complete open-source alternative to LaunchDarkly for teams that need both feature flags and experimentation capabilities. Its feature flag system supports boolean, multivariate, and JSON flags with targeting rules based on user attributes, percentage-based rollouts, and prerequisite flags. The experimentation engine runs statistical analysis using Bayesian, frequentist, sequential, and CUPED methods, giving product teams rigorous A/B testing without third-party analytics dependencies.
What sets GrowthBook apart from simpler feature flag tools is its warehouse-native analytics architecture. Instead of collecting its own event data, it connects directly to your existing data warehouse — BigQuery, Snowflake, Databricks, Redshift, ClickHouse, Mixpanel, or Postgres — to compute experiment results. This means teams use their existing event tracking and data pipelines rather than implementing a parallel analytics stack. The visual experiment editor lets non-technical users create and manage experiments through a web interface.
With 7.9K+ GitHub stars and active development under mixed licensing, GrowthBook has become a default choice for engineering teams that want self-hosted experimentation infrastructure. Most non-enterprise code is available under MIT Expat terms, while enterprise directories use the GrowthBook Enterprise License. SDKs cover JavaScript, React, Python, Ruby, Go, Java, PHP, Swift, Kotlin, Flutter, and more, and the platform includes REST APIs, webhooks, and an MCP Server for feature-flag workflows.
