**[DeepSeek Flash Leaves a 18 Trillion Token Gap, AI Model "Impossible Triangle" Emerges]** Beating AI News Flash: OpenCode CEO Jay V stated that the explosive popularity of DeepSeek Flash has brought immense demand pressure. Over the past two weeks, daily token consumption for this model on OpenCode has surged from approximately 3 trillion to 18 trillion, a sixfold increase. After the price hike, OpenCode has been searching for suppliers capable of handling traffic at prices close to the original rate, but it has been difficult to find providers that simultaneously offer low costs and large-scale capacity. Jay V believes that the challenge DeepSeek Flash faces is not insufficient demand, but rather that the low price has unleashed demand far beyond expectations. It simultaneously delivers near cutting-edge model capabilities and extremely low costs, but once large-scale stable supply is introduced, the AI model market encounters the "Impossible Triangle": affordability, power, and sufficient GPU capacity—three conditions that are hard to meet simultaneously. Following the price increase, OpenCode Go has undergone multiple adjustments. The DeepSeek Flash quota was once reduced from $60 to $15, then restored to $30, and a peak-valley rate mechanism was introduced to align with official pricing. [Original Link]