AI Agent Hub
Back to plugins
🖥️

dsh-fare-meter

Client Updated 2026.08.25

Run the following command in DeepSeek Harness:

dsh plugin install bunipai/dsh-fare-meter

Paste the following prompt into your AI chat to install this plugin:

Run dsh plugin install bunipai/dsh-fare-meter in the DeepSeek Harness terminal to install the plugin, whose source code is at https://github.com/bunipai/dsh-fare-meter ; the fare meter will appear to the left of the model selector after a page refresh.

About this plugin

Most users calling the DeepSeek API only see their total spend on the final bill, with zero visibility into real-time cost during generation. dsh-fare-meter brings a taxi-meter experience into the client: costs tick up by the second while the model is generating and reset to zero the moment it finishes, so you always see what is being spent right now rather than a running total.

A minimal fare capsule (5 decimal places) appears to the left of the model selector. DeepSeek official peak and off-peak schedules are built in (Beijing time 09:00-12:00 and 14:00-18:00 at full rate, all other hours at half), with a clear peak or off-peak label. A V4-Flash official price table is preloaded. Settled usage is computed precisely from session snapshots, while in-flight output is estimated in real time based on measured decoding speed, eliminating the need to look up rate cards manually.

Built for developers and power users who call the DeepSeek API frequently and want immediate cost awareness mid-generation. The plugin does not alter any generation logic; it simply adds a small ticking number to the interface whenever a session is active. Zero intrusion, zero extra configuration.

Use Cases

  • Watch API cost tick up in real time while the model is generating instead of waiting for a bill
  • Pick the cheapest Beijing-time window with peak and off-peak labels right in the UI
  • Estimate in-flight output cost from measured decoding speed to avoid surprise overage

Best For

  • Solo developers making frequent DeepSeek API calls
  • Engineering teams that need tight per-session budget control
  • Heavy users who schedule calls around peak and off-peak windows to cut cost