![[Tool] DeepSeek API Usage Analysis Dashboard — Pure Frontend Multi-Dimensional Visualizer](/_next/image?url=%2Fproject%2Fnextjs%2Fdeepseek-usage-dashboard.png&w=3840&q=75)
[Tool] DeepSeek API Usage Analysis Dashboard — Pure Frontend Multi-Dimensional Visualizer
Why We Built This
As our DeepSeek API usage skyrocketed, the CSV exports from the official dashboard became increasingly painful to decipher. Hundreds of rows of raw numbers — which project is burning the most tokens? Is the cache hit rate where it should be? How are costs distributed across different API Keys? Answers that should be obvious were buried in dull spreadsheets.
To truly understand these key metrics, we decided to build a simple, intuitive, and absolutely secure visualization dashboard ourselves.
Core Features
Four Analysis Dimensions
- Cost Analysis — Total cost, daily cost bar chart, per-key cost distribution donut chart
- Usage Analysis — Total tokens, per-key usage breakdown table, request count statistics
- Cache Analysis — Cache hit rate, daily hit rate trend line chart, per-key hit/miss stacked bar chart
- Trend Analysis — Freely switch between cost / tokens / hit rate / request count dimensions
Practical Highlights
- Smart Multi-Month Merging — Drop billing files from multiple months at once. Automatic filename-based pairing and merging for effortless long-term trend analysis.
- Model Filter — When data contains multiple models, use the Apple-style segmented control to switch views.
- Bilingual UI — Chinese/English interface toggle with dark/light themes. Language and theme preferences are persisted automatically.
Radical Privacy & Modern Architecture
When handling billing data, privacy is a non-negotiable baseline. We chose a pure frontend architecture:
- Framework: Next.js 16 (App Router) + React 19 + TypeScript
- CSV Parsing: Papa Parse (runs in a browser Worker thread, non-blocking UI)
- Chart Rendering: ECharts 5
- Styling: Tailwind CSS + CSS custom-property-driven dual themes
All CSV parsing and chart rendering happens 100% in your browser — no backend servers, no databases, no tracking. Your billing data exists only in your browser memory from start to finish.
About MindRose
This project is developed and open-sourced by Gavin Chen and the MindRose team. MindRose is a tech team focused on delivering lightweight digital solutions for small and mid-sized manufacturers, logistics companies, and cross-border traders.
We don't sell vague "digital transformation" — we use AI and full-stack agile development (Next.js, React, Mendix, and more) to deliver applications that solve real business pain points within weeks. From SEO-optimized trade websites to agile core business systems to AI Agent workflow integration — we understand technology, and we understand your business.
Try It Now
Open https://deepseek-usage.xyz, drag and drop your DeepSeek CSV billing files, and get multi-dimensional usage insights instantly.
Fully open source — stars and contributions are welcome: GitHub Repo