--- title: What X's Open-Sourced Ranking Algorithm Actually Reveals description: X (formerly Twitter) open-sourced the source code behind its ranking algorithm, giving developers and researchers a look at how content is scored and surfaced... canonical: https://daily.dev/posts/what-x-s-open-sourced-ranking-algorithm-actually-reveals-eqkfylyha twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: What X's Open-Sourced Ranking Algorithm Actually Reveals | daily.dev og:description: X (formerly Twitter) open-sourced the source code behind its ranking algorithm, giving developers and researchers a look at how content is scored and surfaced... og:url: https://daily.dev/posts/what-x-s-open-sourced-ranking-algorithm-actually-reveals-eqkfylyha og:image: https://api.daily.dev/og/posts/EQkFYlyHa.png og:image:alt: What X's Open-Sourced Ranking Algorithm Actually Reveals og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # What X's Open-Sourced Ranking Algorithm Actually Reveals **[Collections](https://daily.dev/sources/collections)** · 2 min read · 5 upvotes · 0 comments ## Summary X (formerly Twitter) open-sourced the source code behind its ranking algorithm, giving developers and researchers a look at how content is scored and surfaced in the feed. The release invites analysis of the signals, weights, and heuristics that shape what users see, sparking discussion about transparency in social media recommendation systems. ## Content X released the source code behind its For You timeline and core ranking system on August 13, 2026, under the Apache v2 license. Several people have already dug through the repo (xai-org/x-algorithm), and the picture that emerges is more specific than the usual "the algorithm rewards engagement" hand-waving. Here's what's actually in there. ## How the feed gets built The For You feed pulls from a few different sources, then ranks and filters everything: - **Thunder** — recent posts from accounts you follow, kept in memory for fast retrieval. - **Phoenix Retrieval** — a Grok-based ML model that pulls relevant posts from the wider platform, including accounts you don't follow. - **SimClusters** — surfaces posts from communities that share your interests. All of these get scored by one main model (Phoenix), then filtered by a separate visibility system before anything hits your screen. ## The ranking model Phoenix is a Grok-based transformer that looks at your recent behavior — likes, replies, reposts, clicks, dwell time — and predicts the probability you'll take each possible action on a given post: favorite, reply, repost, quote, share, click, profile visit, and so on. Those probabilities get combined into one score using fixed weights, which are now public. A few numbers stand out: - Predicted quotes carry roughly 5x weight, predicted follows about 4x. - Sharing via DM or copy-link gets a notably large reward — one estimate puts copy-link shares around 20x. - A plain like is worth comparatively little, something like 0.5. - Reports are the strongest negative signal by far, with one figure showing roughly -234. Predicted ## Questions this post answers ### What weights does X's Phoenix ranking model assign to different user actions? Predicted quotes carry roughly 5x weight and predicted follows about 4x, while copy-link or DM shares can be weighted as high as 20x. A plain like is worth comparatively little, around 0.5, and reports are the strongest negative signal by far at roughly -234, far outweighing any positive action. _daily.dev surfaces details like this for engineers reverse-engineering how social ranking systems actually score content._ ### What is the xai-org/x-algorithm repository and what does it contain? It is the open-sourced code behind X's For You timeline and core ranking system, released under the Apache v2 license on August 13, 2026. It includes the Phoenix ranking model, the Thunder retrieval system for followed accounts, Phoenix Retrieval for content from unfollowed accounts, SimClusters for community-based surfacing, and the visibility filtering layer applied after scoring. _engineers studying real-world feed architectures can track releases like this one through daily.dev._ ### How does X's Phoenix model decide what to show in the For You feed? Phoenix, a Grok-based transformer, analyzes recent user behavior such as likes, replies, reposts, clicks, and dwell time to predict the probability of each possible action a user might take on a post. These action probabilities are then combined into a single ranking score using fixed, now-public weights before a separate visibility system filters the results. _anyone building or evaluating ranking systems can follow real algorithm breakdowns like this via daily.dev._ ## Community take How the wider developer community reacted, aggregated from 1 discussion and 2 comments across x (as of 2026-09-13). **TL;DR:** There's barely any substantive discussion here — the few replies are mostly jokes or reactions to the algorithm now being driven by AI-based ranking rather than organic engagement. **Sentiment:** 10% positive · 30% mixed · 60% skeptical **The pushback** - Some feel the shift to AI-driven ranking (Phoenix/Grok-based models) has moved the platform away from simply rewarding good organic content. **By community** - x (mixed): Very light engagement, with one joking reference to AI now dominating what gets seen over a well-crafted post. **Highlights** > @imUrB00gieman so we all must bow to our AI Overlords...I remember when all you had to do is post a 'banger' remember that? > — [GrrrGraphics on x · 3 points](https://x.com/GrrrGraphics/status/2088306610896777414) **Source threads** - [x](https://x.com/imUrB00gieman/status/2088284870304813075) · 9 points · 2 comments --- Tags: [#open-source](https://daily.dev/tags/open-source), [#recommendation-systems](https://daily.dev/tags/recommendation-systems) [View this post on daily.dev](https://daily.dev/posts/what-x-s-open-sourced-ranking-algorithm-actually-reveals-eqkfylyha) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"What X's Open-Sourced Ranking Algorithm Actually Reveals","url":"https://daily.dev/posts/what-x-s-open-sourced-ranking-algorithm-actually-reveals-eqkfylyha","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/what-x-s-open-sourced-ranking-algorithm-actually-reveals-eqkfylyha"},"datePublished":"2026-08-14T16:39:00.774Z","dateModified":"2026-09-13T19:47:06.073Z","description":"X (formerly Twitter) open-sourced the source code behind its ranking algorithm, giving developers and researchers a look at how content is scored and surfaced...","image":"https://pbs.twimg.com/media/HPsL_KRaMAA1JwJ.jpg","thumbnailUrl":"https://pbs.twimg.com/media/HPsL_KRaMAA1JwJ.jpg","isAccessibleForFree":true,"articleSection":"Collections","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Collections","logo":"https://media.daily.dev/image/upload/s--fk_6ycEi--/f_auto,q_auto/v1780996001/logos/collections?_a=BAMAMiWQ0","url":"https://daily.dev/sources/collections"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/what-x-s-open-sourced-ranking-algorithm-actually-reveals-eqkfylyha","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":5},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"open-source,recommendation-systems","timeRequired":"PT2M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Collections","item":"https://daily.dev/sources/collections"},{"@type":"ListItem","position":3,"name":"What X's Open-Sourced Ranking Algorithm Actually Reveals"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/what-x-s-open-sourced-ranking-algorithm-actually-reveals-eqkfylyha#faq","mainEntity":[{"@type":"Question","name":"What weights does X's Phoenix ranking model assign to different user actions?","acceptedAnswer":{"@type":"Answer","text":"Predicted quotes carry roughly 5x weight and predicted follows about 4x, while copy-link or DM shares can be weighted as high as 20x. A plain like is worth comparatively little, around 0.5, and reports are the strongest negative signal by far at roughly -234, far outweighing any positive action. daily.dev surfaces details like this for engineers reverse-engineering how social ranking systems actually score content."}},{"@type":"Question","name":"What is the xai-org/x-algorithm repository and what does it contain?","acceptedAnswer":{"@type":"Answer","text":"It is the open-sourced code behind X's For You timeline and core ranking system, released under the Apache v2 license on August 13, 2026. It includes the Phoenix ranking model, the Thunder retrieval system for followed accounts, Phoenix Retrieval for content from unfollowed accounts, SimClusters for community-based surfacing, and the visibility filtering layer applied after scoring. engineers studying real-world feed architectures can track releases like this one through daily.dev."}},{"@type":"Question","name":"How does X's Phoenix model decide what to show in the For You feed?","acceptedAnswer":{"@type":"Answer","text":"Phoenix, a Grok-based transformer, analyzes recent user behavior such as likes, replies, reposts, clicks, and dwell time to predict the probability of each possible action a user might take on a post. These action probabilities are then combined into a single ranking score using fixed, now-public weights before a separate visibility system filters the results. anyone building or evaluating ranking systems can follow real algorithm breakdowns like this via daily.dev."}}]} ```