Human-like Qwen3.8-27B LoRA gets tool calling in v2.0

On October 3, 2026, a post on Reddit's LocalLLaMA community announced version 2.0 of a LoRA for Qwen3.8-27B that is designed to make the model talk like a person rather than an assistant. According to the author, the update adds tool calling and improves instruction following, two areas where the first release fell short.

The author wrote that the original LoRA was posted a month earlier and drew far more attention than expected: 700+ upvotes, 248 comments and 44k downloads. They said they read every comment, and that users responded well to the tone of voice because they dislike assistant speak. The rest of the feedback, the author said, was critical and fair — the model could not produce more than a few words at a time, had a single default persona, and was completely incapable of using tools. Version 2.0 is presented as closing the tool-calling gap and improving instruction following.

So far the only source for the 2.0 update is the author's own Reddit post. The material contains no independent testing, no benchmark results, and no upvote, comment or download figures for 2.0 itself; those numbers are given only for the earlier version. The post date of October 3 is the only timing detail available for the new release.