ZAI Co Ltd shares climbed sharply on Thursday, rising 9.2% to HK$1,125 after the company publicly confirmed and open-sourced GLM-5.3-Flash. The model, which had been operating in the market under the codename "Ox Alpha" since August 20, was officially unveiled on August 26 with MIT-licensed weights made available on Hugging Face the same day.
GLM-5.3-Flash is described by the company as a 320-billion-parameter mixture-of-experts model that supports multimodal input - including text, images, and video - and offers a 1-million-token context window. The model's release drew immediate attention because of its scale and native multimodal capability, and because it had already been tested publicly while anonymous.
Part of the market response reflected the model's pricing strategy. ZAI priced GLM-5.3-Flash at roughly one-tenth of the rate of the company's flagship GLM-5.3 API, and further reduced the effective cost through a limited-time promotional discount. That aggressive pricing positioned the model as a cost-competitive option among frontier-grade systems and attracted developers who had previously been evaluating it under its stealth identity.
The model first gained traction after its anonymous appearance as Ox Alpha on OpenRouter in August, where observers reported it was matching or outperforming several advanced AI models from peers in the space. Those comparative performance notes contributed to heightened public interest when the company confirmed the model's identity and released the code and weights publicly.
Market sentiment around ZAI's announcement was also buoyed by a favorable backdrop for AI-related equities more broadly. The article notes that strong earnings and guidance from Nvidia helped create a more supportive environment for AI stocks, amplifying the impact of ZAI's product announcement on investor buying.
Context and market reaction
Developers who had access to the model during its anonymous phase appear to have continued testing and evaluating GLM-5.3-Flash after the formal release and open-sourcing. The combination of high-capacity architecture, multimodal capability, public availability of weights under an MIT license, and aggressive pricing drew immediate attention from the developer community and market participants.
The company-specific news occurred against a macro backdrop that the article identifies as supportive for AI equities following a notable earnings and guidance beat from Nvidia, which the article cites as amplifying the catalyst for ZAI's stock move.
Limitations of available information
The article provides a concise account of the model's technical scale, licensing, pricing, and market reaction, but does not provide additional details on long-term commercial plans, revenue implications, or the duration of promotional pricing beyond noting the discount is limited in time.