AI Tech / news
Z.ai’s GLM-5.2 Challenges US AI Giants With Lower Cost and Open Access
Beijing-based startup Z.ai released GLM-5.2, an open-weight LLM that rivals top US models in coding and agent capabilities at a fraction of the cost, sparking comparisons to DeepSeek’s earlier disruption.
Z.ai, a Beijing-based startup, has unveiled GLM-5.2, an open-weight large language model that is drawing comparisons to DeepSeek’s market-shaking debut. The model excels in coding and agentic tasks, requiring minimal prompting to execute complex workflows, and has quickly climbed above Anthropic on the third-party platform OpenRouter.
Former US AI czar David Sacks described GLM-5.2 as “just a tick below Opus 4.8 and right up there with GPT 5.5,” highlighting its competitiveness with premium US offerings. The release comes amid unpredictable US regulations that have throttled models like Anthropic’s Fable and delays in OpenAI’s GPT-5.6 rollout.
GLM-5.2 continues a trend of Chinese AI developers producing powerful yet cost-effective models, challenging the notion that top-tier AI requires billion-dollar budgets. Its open-weight nature allows developers to deploy and fine-tune the model without prohibitive licensing fees.
The rapid adoption on platforms like OpenRouter indicates growing demand among developers for affordable, high-performance alternatives. The international developer community is increasingly looking eastward as reliance on proprietary US models becomes riskier and more expensive.