dsh-conversation-cost

When using DeepSeek Web chat, it is often difficult to intuitively perceive the token consumption and actual cost of a single conversation. This is especially true for mixed-model sessions or scenarios involving large amounts of cached input, where cost calculations can become vague. This plugin aims to provide DSH Web users with a clear, real-time cost monitoring panel.

Its core strength lies in "precise request-by-request billing." Leveraging the registered costLedger on the host side, it folds each request event with O(1) time complexity, avoiding reliance on estimates or log replays. It not only supports independent display of RMB (¥) and USD ($), but also accurately distinguishes between cache hits and misses for input costs. Additionally, it automatically categorizes requests into Off-peak, Peak, or Idle tiers based on the request time, ensuring every penny is transparent and controllable.

This tool is particularly suitable for developers, researchers, and users who need fine-grained control over API usage. If you frequently switch models or handle long conversations, it helps you quickly understand the cost distribution across different tiers via the hover statistics row, allowing you to optimize prompts and model selection to avoid unnecessary expenses.