Early Code Reference Hints at Upcoming Qwen4 Model and 120B Variant

An early code reference discovered in a repository pull request points to the upcoming development of Qwen4. Notably, the reference reveals the existence of a 120B parameter model variant in the pipeline. This early leak provides the local AI community with an exciting glimpse into the next generation of Qwen models. It signals that Alibaba Cloud is actively expanding its high-performance model lineup, which is heavily utilized by open-source developers. The discovery stems from code modifications found in a pull request on GitHub. The inclusion of a 120B parameter configuration suggests a focus on scaling up capabilities for future local deployments.