AI Agent Hub
Back to plugins
🖥️

dsh-model-balance

Client Updated 2026.08.25

Run the following command in DeepSeek Harness:

dsh plugin install jacujay/dsh-model-balance

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

Install from the DSH plugin market or run dsh plugin install jacujay/dsh-model-balance; the project is hosted at https://github.com/jacujay/dsh-model-balance.

About this plugin

Switching between multiple model providers in DSH and wanting to check your balance or quota usually means opening a new tab, logging into each vendor console, and losing your flow context. dsh-model-balance puts that check right into the composer input row: a small diamond icon beside the model name, hover to peek, click to refresh, all without leaving your workflow.

The plugin auto-detects the current default model vendor from DSH configuration (provider id first, baseURL hostname as fallback) and ships built-in support for seven common providers—DeepSeek, Moonshot/Kimi, MiniMax, StepFun, Zhipu GLM, SiliconFlow, and OpenRouter. It handles two result kinds: monetary balance (with granted vs topped-up breakdown where the API offers it) and per-model percentage quotas. For API-compatible proxies like New API, or any other endpoint, the Settings page lets you add a custom balance URL with response field path and auth method. Throughout, API keys are resolved host-side by DSH and never sent to the browser; the local config file stores endpoint metadata only, no secrets.

Built for anyone running DSH web who juggles multiple providers, needs an at-a-glance view of remaining balance or usage, or routes models through a relay gateway.

Use Cases

  • Checking current balance or quota when switching model providers without opening a new browser tab
  • Viewing remaining tokens and usage across multiple models through a New API relay gateway
  • Refreshing balance inside DSH web while API keys stay resolved host-side locally

Best For

  • DSH web developers connecting multiple model providers
  • Users routing model requests through New API or compatible relay proxies
  • Users who prioritize API key security and want keys to never reach the browser