X has now officially open-sourced its recommendation algorithms, making the platform’s main feed-ranking system publicly available for anyone to inspect. In a recent announcement on Musk described the move as a step toward real-time transparency, and said users can now see how X works to fine-tune and refine its recommendation system as changes occur.

X notes that the newly released algorithm is powered by the same Transformer-based architecture used in Grok, an AI model developed by Musk’s artificial intelligence company, XAI. The code has been published on GitHub, allowing developers, researchers, and users to test how content recommendations are generated on X. By open-sourcing the code, the company aims to give developers and researchers direct insight into how content is ranked and recommended on the platform, including both organic posts and ads.

“We know the algorithm is stupid and needs massive improvements, but at least you can watch us struggle to make it better in real time and with transparency,” Musk wrote in his post on X.

What is actually open-sourced

Ax says the open-source codebase includes the logic used to rank and recommend both organic posts and ads on the platform. This means that the systems that determine which posts appear more in a user’s feed, how content is discovered beyond followed accounts, and how engagement signals are weighted are now visible to the public. By open-sourcing the algorithms, X is allowing developers to see how these signals interact, how priorities are determined, and where potential vulnerabilities or biases may exist within the system.

Elon Musk emphasizes transparency for X

Since acquiring X in 2022, Musk has consistently criticized the opaque recommendation systems used by larger social platforms, arguing that users deserve to understand how content is curated. By releasing the code, Musk says he wants to keep transparency as a core principle for X rather than just addressing it through policy statements or blog posts.

The move also follows Musk’s previous announcement of open-source X’s recommendation systems, although this time he claims the release is more complete and reflects the algorithms currently in use. Unlike earlier efforts in 2023, when parts of the Twitter algorithm were shared but later became outdated, this release is being described as an active, production-level system of the platform.