Torvalds claims a 23x performance improvement in a dedicated synthetic benchmark, but he is quick to temper expectations, noting that no real workload will see figures like that, so actual real-world impact will be far less significant. The improvement would plausibly benefit server workloads more than desktop machines.

The patch fits a broadly pro-AI stance Torvalds has taken on LLM tooling within the kernel's tightly controlled, review-heavy environment. Last July he said he was "willing to absolutely put [his] foot down" on AI sceptics, arguing that an automated code review tool used by the kernel team outperforms humans and catches basic mistakes that should never have been submitted. The team is even dropping its requirement that patch authors disclose AI tool use, and Torvalds has likened LLMs to compilers or editors, saying he simply expects people to "do their best and [...] understand what they are doing."

It is not all smooth sailing, though. Because AI agents have become adept at recognising common security flaw patterns, the Linux security mailing list reportedly became "unmanageable" earlier in the year, with the kernel fixing CVEs at a pace of around 2,000 per release. That flood is forcing the team to drop older hardware drivers, which attract many security reports despite their negligible install base.