Zhipu AI has two major developments recently: it released the next-generation GLM-5 model, and it is reportedly planning to list on the Hong Kong Stock Exchange.
GLM-5 Upgrades
Compared to the 4.7 series, GLM-5 improvements mainly include:
- Enhanced reasoning capabilities, with notable gains on math and code benchmarks
- Expanded multimodal capabilities, supporting image-text understanding and generation
- Higher integration of Agent capabilities
Zhipu has not fully disclosed specific benchmark figures yet, but based on available information, GLM-5 appears positioned to directly compete with the latest versions of DeepSeek and Qwen among domestic models.
Hong Kong IPO
If the IPO succeeds, Zhipu will become one of the first batch of large model companies to go public in China.
The timing is notable — Anthropic secured a $38 billion valuation in its Series G round in February this year, and OpenAI also completed a massive funding round last year. AI companies globally are aggressively raising capital. Zhipu's choice to pursue an IPO rather than continued private fundraising may reflect several considerations:
- Secondary market fundraising can reach a broader investor base
- The brand effect of being listed helps secure government and state-owned enterprise contracts
- Early investors need an exit channel
Competitive Positioning
Among domestic large model companies, Zhipu's differentiation has always been academic pedigree plus enterprise services. Born out of Tsinghua University and closely tied to academia, this gives it advantages in technical credibility and talent recruitment.
However, in terms of model capability, DeepSeek and Qwen performed exceptionally well in 2025, and Zhipu's image of "technical leadership" in public perception has somewhat weakened. The dual move of GLM-5 launch plus IPO appears aimed at regaining the narrative.
Sources: Zhipu AI official release, CocoLoop, tech media reports