# szp2005/llm-prices-cn [Health: Active]

**Category:** 🔗 Aggregators  
**Repository:** https://github.com/szp2005/llm-prices-cn  
**GitHub Stars:** 0  
**Views:** 3  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/szp2005-llm-prices-cn

## Description
Daily-verified LLM API pricing dataset (44+ models, CN & global) with a hosted MCP server for live price queries and token cost estimation.

## Claude Desktop Quick Installation
Remote MCP endpoint (confidence: high). Install path detected from listing signals. Add as a URL/SSE server in your client:

```json
"mcpServers": {
  "llm-prices-cn": {
    "url": "https://www.llmabacus.com/api/mcp/mcp"
  }
}
```

## Documentation

## What szp2005/llm-prices-cn MCP server does

The szp2005/llm-prices-cn MCP server makes structured LLM API pricing available to MCP-compatible agents. The underlying dataset tracks 44 models from providers including OpenAI, Anthropic, Google, DeepSeek, Alibaba, ByteDance, Baidu, Tencent, and MiniMax. Pricing is checked daily and includes both Chinese and global vendors.

Model records contain identifiers, display names, vendor and country information, billing currency, input and output prices per million tokens, optional cached-input prices, context-window limits, maximum output lengths, supported modalities, tags, knowledge cutoffs, and quality scores. Prices are represented in both USD and CNY, with an exchange rate recorded in the dataset.

## How it works

The project provides two deployment paths. The hosted MCP endpoint at `https://www.llmabacus.com/api/mcp/mcp` lets agents query pricing remotely and estimate token costs without managing a local dataset. Its documented tools are `query_model_price(model)` and `estimate_cost(text_or_tokens, model)`.

For local use, the server reads the bundled `prices.json` file and communicates over stdio. The local tool set includes `list_llm_prices(vendor?, currency?)` and `estimate_cost(model_id, input_tokens, output_tokens, currency?)`. Because the local server reads its packaged data and does not need network access, it can be used offline after setup.

The repository’s `sync_prices.py` script rebuilds `prices.json` from the source application configuration. The generated file records the last update date, conversion rate, pricing unit, and model array.

## Setup and configuration

To self-host the szp2005/llm-prices-cn MCP server with Docker, build the repository image and run it interactively so the MCP stdio connection remains available:

```bash
docker build -t llm-prices-mcp .
docker run -i --rm llm-prices-mcp
```

A non-Docker setup is also documented. Install the dependencies from `requirements.txt`, then start the server with `python server.py`:

```bash
pip install -r requirements.txt
python server.py
```

No environment variables or API credentials are specified for the local server. To refresh the checked-in snapshot, run `python3 sync_prices.py` from the repository directory.

## Tools and capabilities

The documented MCP operations support these tasks:

- Look up pricing for a selected model through the hosted service.
- Estimate cost from text or token input using the hosted service.
- List local model prices, optionally filtered by vendor or currency.
- Calculate local input and output token costs for a selected model and currency.
- Compare USD and CNY values using the dataset’s recorded conversion rate.

## Limitations and notes

The dataset is a pricing reference, not an API gateway or model execution service. The provided material does not describe tools for sending prompts to model providers. Pricing can change, so applications should account for the dataset’s update date and refresh process rather than treating values as permanent.

The dataset is licensed under CC-BY-4.0. Projects that reuse it must provide appropriate attribution and link back to `https://llmabacus.com`. The documentation does not specify authentication requirements for the hosted endpoint or identify particular MCP desktop clients.

_Full upstream README: https://allmcps.com/mcp/szp2005-llm-prices-cn/readme_