OpenAI and Anthropic Face Threat from Cheaper, Formidable Chinese AI Rival
The Chinese artificial intelligence model GLM-5.2 is attracting significant attention from technology companies and developers in the West. Developed by Beijing startup Z.ai, the model is considered capable of offering abilities close to those of leading products from OpenAI and Anthropic, but at a much lower cost.
Historically, global users have faced a choice between cheaper but less capable Chinese AI models and more powerful but expensive American ones. The arrival of GLM-5.2 appears to be changing this competitive landscape. Launched last month, the model has become a talking point in Silicon Valley for its coding and agentic AI capabilities, with some observers calling its emergence a ‘mini DeepSeek moment’ that could shake up the global AI industry.
The model’s popularity has surged on third-party developer platforms such as OpenRouter, where its usage has surpassed several Anthropic models. Industry figures, including Snowflake CEO Sridhar Ramaswamy and investor Marc Andreessen, have praised its performance. Former Trump AI advisor David Sacks stated that GLM-5.2 is on par with OpenAI’s GPT-5.5 and only slightly behind Anthropic’s Claude Opus 4, warning that the US must not adopt policies that slow its domestic AI development amid intensifying competition.
Analysts suggest that interest in GLM-5.2 has also been fuelled by restrictions on American AI models and delays in the public release of GPT-5.6, prompting developers to seek more open and affordable alternatives. Brian Tse, founder and CEO of Concordia AI, noted that the international developer community is increasingly aware of the risks of relying on proprietary, API-based models from US companies and is considering open-source options for greater flexibility.
Beyond its competitive performance, GLM-5.2’s cost is a major draw. It ranks fifth on the Artificial Analysis language model intelligence index and second on the Code Arena front-end coding leaderboard, yet costs only about one-sixth of premium US models like Claude and GPT. Z.ai founder Tang Jie has stated the company aims to match the latest Anthropic models by the first quarter of next year, though development costs for GLM-5.2 remain undisclosed.
Despite its promise, large-scale adoption by American firms faces significant hurdles due to data security concerns, particularly in heavily regulated sectors such as banking and cybersecurity. Counterpoint Research analyst Wei Sun noted that many European and US companies remain reluctant to integrate Chinese AI models into their technology infrastructure, with some clients and regulators likely to reject them regardless of performance or cost efficiency.
A RAND report indicated that the global market share of Chinese large language models jumped from 3% to 13% within two months of DeepSeek’s R1 launch in January last year, with the highest adoption in developing nations and countries with close economic or political ties to Beijing. While large corporations may move slowly, experts believe startups and small businesses are quicker to adopt Chinese AI models. Technology analyst Poe Zhao noted that developers tend to prioritise performance, cost, and ease of access over the model’s country of origin. He suggested that while GLM-5.2 may not fully replace OpenAI or Anthropic, it could become a significant alternative in the global AI ecosystem, especially for those seeking low-cost solutions with near-leading capabilities.