@xenovacom posted an announcement introducing Fleet, described as GPU benchmarking entirely in the browser. The post states that users can run WebGPU compute kernels drawn from real AI workloads directly in that environment. Georgi Gerganov, creator of llama.cpp, retweeted the message. A generated summary in the packet repeats that Fleet enables browser-based GPU benchmarking using those kernels. The packet contains no independent confirmations, launch dates, or additional details beyond the original post and retweet.