Qwen/Qwen3-235B-A22B · Two of the base models are missing

Two of the base models are missing:

Qwen3-235B-A22B-Base
Qwen3-32B-Base

Since the blog post announcing Qwen3 release mentioned Qwen3-235B-A22B-Base's results on various benchmarks, I sincerely urge that the two base models be released.

This is very beneficial for these aspects:

Reproducibility of benchmark results
Validation of scaling law
Fine-tuning on a different domain from math, reasoning and code, such as novels

The exclusion of Qwen3-32B-Base and Qwen3-235B-A22B-Base makes the release feel inconsistent. A complete and consistent release would both express sincerity and better benefit the community.