Rumored Midsize Qwen Open-Weight Model Expected to Launch Next Week
A Qwen community manager on Discord teased the upcoming release of a new midsize open-weight model, potentially from the Qwen 3 family, expected as early as next week. The manager noted that there will be no early access for this model due to scheduling constraints. Qwen is a highly influential open-weight model family developed by Alibaba, and new releases often set new performance benchmarks for the open-source AI community. A new midsize model could provide developers with a powerful, cost-effective alternative for local deployment. While the community manager reacted negatively to a user asking for a 35B parameter model, some community members speculate the new model could be larger, potentially exceeding 100B parameters. However, because this information comes from a Discord teaser, official technical specifications and release dates remain unconfirmed.
## BACKGROUND
Developed by Alibaba Cloud, Qwen is a prominent family of large language models that includes open-weight versions like QwQ-32B-Preview. Open-weight models make their trained parameters publicly downloadable, allowing developers to run and fine-tune them locally, unlike fully proprietary models.