# PromptSpend

> PromptSpend is an open-source MCP server for ai models and llm tooling. LLM price lookups with source and last-verified date, plus workload cost estimates. It can run locally on your machine or as a hosted remote server.

- **Kind:** MCP server
- **Pricing:** Open source
- **Platforms:** CLI, API
- **Website:** https://promptspend.com
- **Repository:** https://github.com/andrewavery7/promptspend/tree/HEAD/mcp
- **MCP Registry name:** io.github.AndrewAvery7/promptspend
- **Categories:** AI models and LLM tooling, finance and payments
- **Listing updated:** 2026-11-14

## About

Most price lists give a number and nothing else. This server attaches a provenance block to each rate: where it came from (such as the provider's own pricing page), the date it was last confirmed and a URL to check. When two sources disagree and nobody has adjudicated, the response is marked disputed and shows both figures instead of quietly choosing one.

Its estimate_cost tool runs the same cost engine as the promptspend.com website, imported rather than reimplemented, and a test asserts that the two agree. The README says it accounts for conversation history re-sent each turn, cache writes costing more than input, per-request long-context tiers, billable reasoning tokens and separately priced output.

The author also budgets for the server's own weight: three tools with short descriptions, about 897 tokens added per turn, checked in CI. It installs with npx, needs no API key or account and is MIT-licensed, and a remote endpoint is also listed.

## Features

- **Sourced prices:** get_price returns input and output rates per million tokens with source, date and URL.
- **Disputed flag:** Conflicting sources are marked disputed with both figures shown.
- **Workload costing:** estimate_cost prices a described workload, not just a per-token rate.
- **Shared cost engine:** Uses the website's own engine so MCP and web numbers cannot drift.
- **Small footprint:** Three tools, about 897 tokens per turn, checked in CI.
- **No account needed:** MIT licensed and runs without an API key.

## Good for

- Compare what a support assistant costs on two different models
- Cite a model's price with the date it was verified
- Spot prices that sources disagree about before budgeting
- Estimate monthly spend for a chat product with growing history

## Install

### Claude Code (Terminal)

```sh
claude mcp add --transport http promptspend https://promptspend.dev/mcp
```

### Claude Desktop (Settings → Connectors → Add custom connector)

```sh
https://promptspend.dev/mcp
```

### Cursor (~/.cursor/mcp.json)

```json
{
  "mcpServers": {
    "promptspend": {
      "url": "https://promptspend.dev/mcp"
    }
  }
}
```

### VS Code (.vscode/mcp.json)

```json
{
  "servers": {
    "promptspend": {
      "type": "http",
      "url": "https://promptspend.dev/mcp"
    }
  }
}
```

### Windsurf (~/.codeium/windsurf/mcp_config.json)

```json
{
  "mcpServers": {
    "promptspend": {
      "serverUrl": "https://promptspend.dev/mcp"
    }
  }
}
```

### Other clients (your client's MCP settings (most use this shape))

```json
{
  "mcpServers": {
    "promptspend": {
      "type": "http",
      "url": "https://promptspend.dev/mcp"
    }
  }
}
```

## Tools

- `get_price`: Input and output price per million tokens for a model, with provenance.
- `estimate_cost`: Estimate the cost of a described workload using the promptspend.com cost engine.

## FAQ

### Do I need an API key?

No. The README says no API key and no account, and it is MIT licensed.

### How do I install it?

Run claude mcp add promptspend -- npx -y @promptspend/mcp.

### How many tokens does it cost my context?

The README puts it at about 897 tokens per turn, with a 1,200 budget enforced in CI.

### What does estimate_cost account for?

History growth, cache write premiums, long-context tiers, reasoning tokens and separately priced output.

### Is PromptSpend free?

Yes. PromptSpend is open source, so you can use and self-host it for free.

### How do I add PromptSpend to Claude Code?

Run this command in your terminal: claude mcp add --transport http promptspend https://promptspend.dev/mcp

### What are the best alternatives to PromptSpend?

The closest alternatives to PromptSpend are Claude Code Ultimate Guide, Solana MCP by Vybe, Tapetide - Indian Stock Market MCP, Budget Tracker and Quantivex Research.

## Alternatives

- [Claude Code Ultimate Guide](https://ai-powered.com/mcp/claude-code-ultimate-guide): Claude Code Ultimate Guide is an open-source MCP server for ai models and llm tooling.
- [Solana MCP by Vybe](https://ai-powered.com/mcp/solana-mcp-by-vybe): Solana MCP by Vybe is an MCP server for finance and payments.
- [Tapetide - Indian Stock Market MCP](https://ai-powered.com/mcp/tapetide-indian-stock-market-mcp): Tapetide - Indian Stock Market MCP is a freemium MCP server for finance and payments.
- [Budget Tracker](https://ai-powered.com/mcp/budget-tracker-2): Budget Tracker is a freemium MCP server for finance and payments.
- [Quantivex Research](https://ai-powered.com/mcp/quantivex-research): Quantivex Research is a free MCP server for finance and payments.
- [Provyan](https://ai-powered.com/mcp/provyan): Provyan is a free MCP server for finance and payments.
- [TribuJus](https://ai-powered.com/mcp/tribujus): TribuJus is an open-source MCP server for legal and government.
- [L2WS](https://ai-powered.com/mcp/l2ws): L2WS is a paid MCP server for finance and payments.

---
Source: https://ai-powered.com/mcp/promptspend
