The race to develop ultra-low-cost AI models is heating up, with two major players, (Zhipu) and DeepSeek, emerging as frontrunners in the market. Recent developments have confirmed that Zhipu’s “Ox Alpha” model, now rebranded as GLM-5.3-Flash, is gaining traction with its pricing strategy of $0.15 per million input tokens and $0.50 per million output tokens. This has sparked a crucial read-through for the industry – falling inference costs are not only making AI more accessible but also unlocking substantially more usage, particularly in the Chinese market.

The growing demand for AI models is translating into increased compute demand, with more businesses and organizations looking to incorporate AI into their operations. This trend has significant implications for the hardware debate, as cheaper intelligence leads to more aggregate compute being used. The race to develop ultra-low-cost AI models is far from over, and it will be exciting to see how Zhipu and DeepSeek continue to innovate and adapt their strategies in response to changing market dynamics.