[Zhitu API Users Near 7 Million, Over 50,000 Domestic AI Chips Deployed to Ease Computing Power Pressure] BlockBeats News, August 10, according to a report by LatePost, the registered user base of Zhitu MaaS has approached 7 million, an increase of about 2 million since early July, with 23,000 enterprise clients. Its developer product, ZCode, surpassed 1 million users within a month of launch. The report states that Zhitu's ARR (Annual Recurring Revenue) has grown approximately 15-fold this year. Market insiders estimate that Zhitu's current ARR may have reached $2 billion, though this figure has not been officially confirmed. As demand for model calls grows rapidly, Zhitu is expanding its domestic computing power infrastructure. It is reported that Zhitu has deployed over 50,000 domestic AI computing chips to address the increasing inference demand. Previously, market rumors suggested that Zhitu had built domestic AI computing infrastructure with a capacity of up to 1GW. Zhitu's growth has been primarily driven by the surge in demand for programming scenarios. Following the release of the GLM-5 model, the company's ARR grew rapidly, with its API business now becoming the core revenue source. Zhitu is also continuously optimizing inference efficiency, improving computing power utilization through technologies such as KV cache partitioning and inference network architecture. Additionally, Zhitu launched a high-speed API in July, capable of generating 400 tokens per second, approximately 8 times faster than the standard API. On July 31, the company reopened the purchase of its Coding Plan and significantly adjusted its pricing, raising the monthly fee for the Lite version to 118 yuan. Industry experts believe that as demand for AI programming continues to grow, optimizing inference efficiency will become a key competitive advantage for large model companies. Those who can generate more tokens with the same computing power will gain an edge in cost and commercialization. Currently, Chinese large model companies such as Zhitu, Kimi, and DeepSeek are all facing computing power bottlenecks and competitive pressure from models. The market expects several companies to launch next-generation models in August, further intensifying competition in the AI large model space. [Original Link]