The big model price war we talked about a week ago has really come. Today, several AI big model companies coincidentally did the same thing: 1) Anthropic announces the reset of 5-hour and weekly usage restrictions for all users; 2) OpenAI has once again increased the usage quota for Codex and ChatGPT Work. Pay attention to the word "again" to indicate that it is not the first time restrictions have been lifted, but rather multiple rounds of recent upward adjustments; 3) XAI not only reset the usage limit of Grok, but also opened Grok Build; 4) Cursor is more direct, doubling the model usage limit for all subscription plans. Seeing such actions, people are most concerned about two questions: why do we do it this way? What is the impact? 1. Why do we have to do this first? Is the core still the https://(x.com)/qinba frank/status/2076839550119444702 we talked about together these two weeks? S=46&t=k6rimWSEbo2D2TXolYcM-A Enterprises adopt AI to enter the engineering stage of segmented scenarios and multiple models. Enterprises themselves also need to allocate high-frequency, low complexity, and low value scenarios to high cost-effective tokens, which forces them to actively compete with cutting-edge closed source large models. Price war is always the most effective strategy. 2. What other impact does it have? 1) In the short term, people are concerned about whether it will affect the profit margin of the large model. After all, a decrease in price means a decrease in single call revenue, a decrease in the commercial value of a unit token, and a natural pressure on short-term gross profit margins and profitability. 2) But in the medium to long term, it depends on how much demand growth can be stimulated. If the explosive growth of demand and the significant increase in token consumption offset the decrease in token unit price, the total plate will eventually become larger. Last week here https://(x.com)/qinba frank/status/2074061695438483754? S=46&t=k6rimWSEbo2D2TXolYcM-A has also been discussed: "The more companies adopt AI, the better the cost engineering, and the market share of high-end models is partially eroded. However, will it drive the absolute value of commercial market share to continue to grow rapidly? Because the plate is bigger. ” But the most crucial thing here is how big the time difference is between the short and medium to long term. The profit margin and profitability of price reduction will immediately decrease. Can the explosion of demand be compensated for immediately, or will it take some time?