# Chinese firms release open-weight AI models

> Live situation record from CLSTR: https://clstr.news/situations/chinese-firms-release-open-weight-ai-models
> Updated: 2026-09-09T13:45:52.000Z. Sources: 26. Developments: 2.

Chinese technology firms have released several new open-weight AI models utilizing Mixture-of-Experts (MoE) architectures to improve efficiency.

In late August, Z. ai released GLM-5.3-Flash, a multimodal model with 320 billion total parameters designed for text, image, and video inputs. Concurrently, Alibaba introduced Qwen3.8-Flash-Next, a 125-billion-parameter model optimized for long-context processing and tasks such as agentic coding.

Following these releases, Ant Group introduced Ling-3.0-flash-Fin, an open-weight model specifically engineered for professional financial workflows. This 124-billion-parameter MoE model focuses on specialized capabilities including valuation modeling, authoritative information retrieval, and structured report generation.

## Timeline

### 2026-09-09: Ant Group releases Ling-3.0-flash-Fin open-weight financial AI model

Ant Group has open-sourced Ling-3.0-flash-Fin, an open-weight AI model using MoE architecture to automate financial research, valuation modeling, and reporting.

2 sources. https://clstr.news/cluster/ant-group-releases-ling-30-flash-fin-open-weight-financial-ai-model

### 2026-08-25: Z. ai and Alibaba release new efficient open-weight AI models

Z. ai released the open-weight GLM-5. 3-Flash multimodal model, while Alibaba debuted Qwen3. 8-Flash-Next as a preview of its Qwen4 architecture, both focusing on parameter efficiency and long-context tasks.

24 sources. https://clstr.news/cluster/alibaba-to-release-qwen-38-flash-next-as-preview-of-qwen-4

---
Cite as: Chinese firms release open-weight AI models. CLSTR, https://clstr.news/situations/chinese-firms-release-open-weight-ai-models