X Announces Open Source Release of Recommendation System
At 07:30 on August 14, 2026, X—formerly Twitter—made the source code for its "For You" feed publicly available. The release is hosted on GitHub under the Apache v2 license. Users can now access model configurations, filters, and signal-weighting parameters, which X's Vice President of Product, Keith Coleman, says should lead to a clearer understanding of how the system operates. This transparency push comes as social platforms increasingly face scrutiny over their algorithmic decision-making.
"You'll get the core ranking code that pulls posts, ranks them for any given user, and assembles the feed." Keith Coleman
Additionally, X is introducing a new tool called Under the Hood, designed to show users how ranking algorithms impact their account. To receive statistics, an account must have at least 10 posts from the previous month, and the results can be downloaded as a JSON file. The tool will initially roll out to a test group of accounts that have existed for at least one year.
"Our dream is that anyone can assess how posts are distributed on the platform, verify that the playing field is level, and if they believe otherwise—criticize it so we can continue to improve and fix the system." Keith Coleman
Before making the code public, X demonstrated it to external researchers, who launched the evaluation system called Phoenix. Notably, the systems that use Grok to anticipate rule-breaking posts were not included. Developers can also submit questions, and X engineers will consider them for potential revisions. Prior to Musk's purchase, Republican members of the U.S. Congress had accused Twitter of shadow-banning their content. The recommendation codebase has grown by roughly 10 to 15 times, underscoring the scale of changes on the platform.
Future Prospects and Social Media Impact
By releasing this code and offering the new user-facing tool, X is underscoring its stated goal of transparency and stronger user trust. It could also represent a notable step for how platforms interact with their audiences, especially amid ongoing criticism over content management.
Similar initiatives might inspire other social networks to pursue comparable transparency measures, shaping a broader industry trend toward opening up algorithms in media and communications.