LinkedIn is rebuilding its main feed algorithm with a new ranking system powered by advanced large language models (LLMs) and graphics processing units (GPUs), according to a recent company announcement.
The update aims to give the platform a broader view of user behaviour and respond to members’ evolving interests. LinkedIn said the new system is designed to better understand what posts are about and how they relate to a user’s activity on the platform, with the goal of delivering more accurate and useful recommendations.
Traditionally, LinkedIn’s feed algorithm has relied on signals such as a user’s industry, experience, skills and location, along with engagement history, to recommend content relevant to professional interests and goals.
With the new approach, the Microsoft-owned platform has combined several retrieval sources, including trending content and collaborative filtering, into a single system. LinkedIn said this replaces its previous ranking model, which evaluated each feed impression independently and did not account for patterns in how professionals consume content over time.
The new system uses LLMs to create more detailed representations of users and content, while a GPU-powered ranking model is designed to capture how members engage with posts across multiple interactions.
LinkedIn said the update is intended to give creators more opportunities to reach a wider range of users while increasing the visibility of newer posts in the feed.
According to the company, the system will continuously update its understanding of both content and user interests, which should make the feed more responsive to changes in behaviour.
“When industry news breaks and relevant posts start getting traction, you see them within minutes, not hours,” the company said in its announcement. “When you engage with content signaling a new professional interest, subsequent Feed visits reflect that updated understanding almost immediately.”
The changes come less than a year after LinkedIn reversed an earlier algorithm update that had reportedly resurfaced older posts in users’ feeds.
Over the coming months, LinkedIn said it will continue refining the system with the aim of reducing repetitive, click-driven posts and filtering out engagement bait while prioritising relevant content.
