Why is this happening? On one hand, computational power optimization and model compression have really lowered costs; on the other hand, it's outright competition—whoever is cheaper gets to attract users first, build ecosystems, and gather data. Especially in the Chinese market, the competition has been fierce for a few years now, with lightweight models being offered for free and high-performance ones dropping dramatically, even causing profit margins to turn negative at times.

However, this battle has now reached a point where it’s no longer about endlessly cutting prices. Some companies are starting to stabilize or even slightly raise prices, shifting toward tiered pricing, bulk discounts, and cache optimization—more refined strategies. After all, burning money endlessly isn’t a sustainable long-term plan; everyone wants to grow their scale first and then gradually turn a profit. This is good news for our users! AI is becoming cheaper and more enjoyable to use, and developers can boldly try new applications. But for companies, the pressure is immense—they have to compete on technology, efficiency, and ecosystems; whoever falls behind will be left out.