dsh-cost-estimate

One of the biggest pain points when using DeepSeek for deep conversations or processing long documents is the unpredictability of costs. When you send a complex, long question to the model, you often have to wait until the answer is complete to see the actual Token usage and cost. This "hindsight" experience makes it difficult to control budgets in real-time.

dsh-cost-estimate is designed to solve this problem. It achieves a "quote before answering, check bill after" workflow by inserting a real-time inline notification into the Web chat stream. Before the model starts generating, it intelligently predicts the input and output Token ranges and estimated costs based on historical data; once the answer is finished, the same line automatically flips to display the actual consumption.

Beyond the intuitive preview, this plugin features advanced capabilities such as peak/off-peak pricing strategies based on Beijing time and a cross-session calibration mechanism that refines estimates over time. It is perfect for developers who need to iterate on complex Prompts or researchers handling large volumes of text, allowing them to clearly grasp the economic cost of each turn and achieve precise budget control.