Recently, a subtle trend has emerged in the global AI application market. As several leading US AI companies continue to adjust their pricing strategies, many US enterprises have begun actively seeking more cost-effective alternatives. Chinese large models, with their excellent cost-performance ratio, are accelerating their integration into the application workflows of US enterprises.

According to the latest industry data, since February 8th this year, the proportion of US enterprises calling Chinese mainstream models such as DeepSeek and Zhipu GLM has significantly increased. In the statistics of the AI model aggregation platform OpenRouter, these domestic models have maintained a weekly call share of over 30%, with peaks reaching up to 46%. Compared to data from the past year, this growth has been very rapid.

Industry insiders analyze that this shift is driven by strict cost control considerations by enterprises. When tasks do not require so-called "strongest" models, companies tend to route to models that are "good enough and low-cost." According to relevant data, domestic open-source and open-weight models have shown strong cost competitiveness, with prices generally 60% to 90% lower than top-tier US models.

In addition to price advantages, the catching up of technical capabilities is also a key factor for the breakthrough of domestic models. Currently, the performance gap between top domestic models and leading US models has narrowed to 6 to 9 months. In practical application benchmark tests, some Chinese models show only minor differences in specific tasks compared to US leading models, but at only one-fifth the cost.

Many startups have already begun to achieve cost reduction and efficiency improvement by switching models. Industry cases show that by migrating workflows from high-cost models to high-value domestic models, companies can save millions of dollars in operating expenses within a few months. This competitive strategy based on performance and cost is giving Chinese AI models unprecedented vitality in the international application market.