- Chinese AI models are gaining traction in the US market, with z.AI's GLM-5.2 outperforming GPT-5.5 at roughly one-sixth the cost.
- z.AI released GLM-5.2 on June 16 as an open-source coding model with a 1 million token context window; independent reporting confirms it outperforms GPT-5.5 on SWE-bench Pro (62.1% vs 58.6%) at roughly one-sixth the cost.
- Per DataCamp and independent practitioner testing, GLM-5.2 also outperforms GPT-5.5 on Terminal-Bench 2.1 (81.0), at roughly one-sixth the API cost.
- The model ships with a 1 million token context window, up to 131,072 output tokens, MIT-licensed open weights, and an Anthropic-compatible endpoint that works with tools like Claude Code and Cline without new integrations.
Chinese AI models are increasingly appealing to U.S. customers, with z.AI's GLM-5.2 outperforming GPT-5.5 on benchmarks while costing significantly less. Released on June 16, GLM-5.2 boasts a 1 million token context window and has shown superior performance in tests, achieving 62.1% on SWE-bench Pro compared to GPT-5.5's 58.6%.124
The New York Post reports that this trend has raised concerns regarding the data sourcing practices of Chinese companies, with allegations of using distillation techniques to harvest proprietary AI data. These claims have been echoed by U.S. firms, including OpenAI and Anthropic, who have publicly raised similar concerns.
Despite these accusations, the Chinese Embassy has labeled them as “groundless.” Meanwhile, industry leaders like Elon Musk predict that China will achieve frontier-class AI by Q1 2027, although z.AI co-founder Jie Tang believes it will happen sooner.
In a related development, Microsoft is considering a self-hosted version of DeepSeek-V4 to enhance its Copilot Cowork enterprise tool, aiming to reduce costs while providing customers with options to select their preferred AI model.
“Chinese AI models are increasingly appealing to US customers, with z.AI's GLM-5.2 outperforming GPT-5.5 on benchmarks. This shift highlights the competitive landscape in AI technology and pricing.”
Let's Data Science