# WeChat Begins Limited Beta Test of Native AI Assistant 'Xiao Wei'

- Category: Flash (快讯)
- Published: 2026-06-22T12:00:00.000Z
- Source date: 2026-06-22
- HTML: https://www.88z.ai/en/news/flash-20260622-8f34258e2417f8a1
- Markdown: https://www.88z.ai/en/news/flash-20260622-8f34258e2417f8a1.md
- Other language: https://www.88z.ai/news/flash-20260622-8f34258e2417f8a1.md

WeChat has launched a limited beta test for its native AI assistant, 'Xiao Wei,' accessible via a green eye icon in the top-left corner of the main interface. The assistant supports text and voice interactions to perform native functions such as adjusting settings, sending messages, launching mini-programs, and generating images. It also features the ability to create mini-programs from a single sentence. The core model is WeChat's self-developed WeLM, with some responses powered by DeepSeek.

## At a glance

- WeChat's native AI assistant 'Xiao Wei' enters beta, supporting voice and text control for settings and messaging
- Powered by self-developed WeLM with DeepSeek integration, featuring one-sentence mini-program generation

## The update

WeChat has officially started a limited beta test for its native AI assistant, 'Xiao Wei.' Users can access the service by clicking the green eye icon located in the top-left corner of the main interface. Xiao Wei supports text or voice conversations to operate various native WeChat functions, including adjusting system settings, sending chat messages, quickly launching mini-programs, and generating images. Additionally, the assistant features an innovative capability to create mini-programs directly from a single-sentence command. On the technical side, the primary driving model for Xiao Wei is WeChat's self-developed WeLM, while the DeepSeek model is called upon for certain responses to enhance service capabilities.

## Sources

- [财联社](https://mp.weixin.qq.com/s/nXjFyV79SEM5l7J7dqBkgQ) (2026-06-22)

This article is an AI-assisted synthesis by 88Z AI based on the public sources below. Product claims and availability are attributed to their publishers. It does not represent independent testing or first-hand reporting.