Hikvision updates Guanlan video AI

Hikvision said on September 8 that it had refreshed its Guanlan AI applications, pitching higher accuracy, lower operating cost and more natural interaction across recorded and live video. The Hangzhou-based company said the update builds on Guanlan, its large-scale AI model suite launched more than a year ago. In the company’s description, the latest applications are designed to let video systems identify what is in a scene, work across previously recorded footage and respond to user requests in natural language. ### What did Hikvision say changed in Guanlan? Hikvision said the updated applications are centered on three points: “higher accuracy,” “lower cost” and “smarter interaction.” The company said those gains apply to AI detection and analysis, as well as to the way users search and work with video. The September 8 announcement said current video systems can “understand” what they are looking at, know what has already been recorded and, through natural-language features, recognize what a user wants. (prnewswire.com) Hikvision said those capabilities are meant to reduce manual work and make operations safer and cheaper to run. ### How is the company describing natural-language video search? (prnewswire.com) Hikvision said natural-language innovations are a core part of the refresh. The company’s wording indicates users can query both archived and live video without relying only on fixed menus, tags or manually built search rules. The release said the system can reason over what has already been recorded, alongside what is happening in current video streams. (prnewswire.com) Hikvision did not, in the materials reviewed, disclose benchmark figures, pricing changes or named customers using the updated Guanlan applications. ### Where does this fit inside Hikvision’s broader AI lineup? (prnewswire.com) Hikvision describes Guanlan as its proprietary suite of large-scale AI models for its AIoT ecosystem. On its product pages, the company says Guanlan is intended to support applications at the system edge and across industry deployments that rely on video, sensing and automation. The September 8 release ties that broader model platform to specific application behavior inside video systems. (prnewswire.com) Rather than presenting Guanlan only as a base model family, the company framed the update around usable functions such as detection, analysis and natural-language interaction with video content. ### What details are still missing from the announcement? (hikvision.com) Hikvision’s statement did not provide independent performance tests, side-by-side comparisons with earlier Guanlan releases or a timetable for regional rollout. The announcement also did not specify whether the lower-cost claim comes from reduced compute use, fewer false alarms, lower storage needs or less human review. (prnewswire.com) PR Newswire listed the announcement at 11:48 p.m. Eastern on September 7, corresponding to September 8 in Hangzhou, China. Hikvision’s release says the updated Guanlan AI applications are available through the company’s product and solutions channels, where it is also publishing material on the Guanlan model family. ### What comes next for users watching this rollout? (prnewswire.com) Hikvision said on September 8 that the next reference point for customers is its product documentation and solution pages covering Guanlan applications and the broader Guanlan large-scale AI models. The company has not, in the announcement reviewed, set a separate future event, launch date or pricing milestone beyond the release itself. (prnewswire.com 1) (prnewswire.com 2)