- Published on July 30, 2025
- In Global Tech
A New Open Source Model From China is Crushing the Benchmarks
And in some situations, it is even better than Anthropic’s Claude 4 and OpenAI’s o3.
- By Supreeth Koundinya
Ziphu AI — also known as Z.ai — has unveiled the new GLM 4.5 family of AI models, which the company claims outperform Anthropic’s highly regarded Opus 4.0 on several benchmarks. The release follows Moonshot AI’s Kimi K2, which exhibited benchmark performance.
Z.ai, based out of Beijing and backed by the e-commerce giant Alibaba, released the GLM-4.5 and the GLM-4.5-Air AI models.
Looking at the recent releases of open source AI models from the East, it wouldn't be an overstatement to say that the future of open source AI may not be led by the West. “It seems like Chinese labs are playing musical chairs at this point,” said Satvik Paramkusham, an engineer on X.
https://twitter.com/Yuchenj_UW/status/1950034092457939072
Both – the GLM-4.5 and the
📣 Want to advertise in AIM? Book here
Supreeth Koundinya
Supreeth is an engineering graduate who is curious about the world of artificial intelligence and loves to write stories on how it is solving problems and shaping the future of humanity.
Related Posts
Mangaluru Looks to Build Its Own Tech Identity, Not Replicate Bangalore
“The coastal city could showcase tangible results by applying deep tech to areas it already dominates”
Google’s Gemini Nano Banana and the Cost of Convenience
The company’s new AI image and photo editor deepens concerns over data use and consent gaps, experts warn.
BharatGen’s ‘Recipe’ for Building a Trillion Parameters Indic Model
The consortium insists sovereignty doesn’t mean shutting the door on global players.
Storytelling is in the Creator’s Control with GenAI
“Democracy would not have been possible without storytelling being distributed.”
India’s Gas-Powered Data Centres at Crossroads: Bridge Fuel or Wrong Turn?
Natural gas may seem a potential fuel for data centres, but higher costs and insufficient infrastructure pose challenges.
How BharatGen Took the Biggest Slice of IndiaAI’s GPU Cake
BharatGen has secured 13,640 H100 GPUs and ₹988.6 crore in funding to pursue India’s first trillion-parameter AI model initiative.
The Generation That Refused to Log Off in Nepal
“The backlash made it evident that Nepali citizens do not tolerate digital authoritarianism disguised as governance.”
Should India Build Its Own AI Foundational Models?
At Cypher 2025, industry leaders debate whether India should invest in building its own AI foundational models or adapt global
Download the easiest way to
stay informed
stay informed
