# XF Model Source

> XF Model Source is a freemium MCP server for ai models and llm tooling. Hosted MCP server that shortlists and tests AI models against your own job. It runs as a hosted remote server, so there is nothing to install locally.

- **Kind:** MCP server
- **Pricing:** Freemium
- **Starting price:** $19 a month
- **Platforms:** API
- **Website:** https://xpansion.dev/xfms
- **Repository:** https://github.com/visionairyse/xfms
- **MCP Registry name:** dev.xpansion/xfms
- **Categories:** AI models and LLM tooling
- **Listing updated:** 2026-11-15

## About

Instead of picking from a general leaderboard, XF Model Source starts by interviewing you. It asks a few plain questions about what goes in, what must come back, the output format, the cost ceiling and what must never happen, then turns the answers into requirements for the job.

It compares 450+ OpenRouter models on more than 200 attributes, drawing on independent test scores, published capabilities and prices, and its own comparison results. Where a requirement cannot be measured or inferred, the model is shown as unknown rather than good. A screening round then runs about 12 likely models on one of your sample tasks, drops any that break a must-have, and runs the best six on all your tasks.

The server is hosted at xfms.xpansion.dev, so there is nothing to install. Sign-in is by email, and comparisons use your own OpenRouter key, passed as an X-OpenRouter-Key header and not stored.

## Features

- **Job requirement interview:** Asks only the questions that change the answer and writes down the job's real requirements.
- **Wide model comparison:** Rates 450+ OpenRouter models across more than 200 attributes, marking unknowns as unknown.
- **Screening round:** Runs about 12 candidate models on one sample task and removes any that break a must-have.
- **Full test on your tasks:** The best six run all your tasks, and you get the winner, the reasons and every answer.
- **Cost cap before tests:** Shows the most a test can cost before anything runs.
- **XF Model Watch add-on:** Optional weekly re-tests of saved jobs with email alerts and retirement warnings.

## Good for

- Pick the cheapest model that extracts invoice fields into JSON correctly
- Choose a model for answering support emails without inventing policies
- Re-check a saved model choice when prices or model lineups change
- Compare shortlisted models on your own sample tasks before committing

## Install

### Claude Code (Terminal)

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

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

```sh
https://xfms.xpansion.dev/mcp
```

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

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

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

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

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

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

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

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

## Tools

- `discover`: Finds the job's real requirements.
- `rank`: Returns the full shortlist with the evidence behind it.
- `pick`: Returns the single best proven model.
- `compare`: Runs the screening round and test on your tasks.
- `benchmark`: Benchmarks models.
- `field_results`: Shows what has worked for jobs of the same kind.
- `watch_job`: Sets up XF Model Watch for a job.

## FAQ

### Does it cost anything to use?

Discovery and the shortlist are free, and sign-in needs only an email. Tests run on your own OpenRouter key, usually a few cents each.

### Which key do the tests use?

Your own OpenRouter key, sent as an X-OpenRouter-Key header. The README says it is used for that run and never stored.

### What does XF Model Watch cost?

It is optional and starts at $19 a month for 3 jobs; Team is $49 for 10 jobs and Agency is $149 for 40, with a 14-day money-back.

### Which assistants can connect?

Claude, Claude Code, Codex, Cursor, Windsurf and other clients that support HTTP MCP servers.

### Is XF Model Source free?

XF Model Source has a free plan, with paid plans for more usage or features. Paid plans start at $19 a month.

### How do I add XF Model Source to Claude Code?

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

### What are the best alternatives to XF Model Source?

The closest alternatives to XF Model Source are Claude Code Ultimate Guide, signafAIde, Toolkien, Lugh and HypePaper.

## 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.
- [signafAIde](https://ai-powered.com/mcp/signafaide): signafAIde is a freemium MCP server for security.
- [Toolkien](https://ai-powered.com/mcp/toolkien): Toolkien is an open-source MCP server for ai models and llm tooling.
- [Lugh](https://ai-powered.com/mcp/lugh): Lugh is an open-source MCP server for ai models and llm tooling.
- [HypePaper](https://ai-powered.com/mcp/hypepaper): HypePaper is a freemium MCP server for ai models and llm tooling.
- [Agent Data Tools: Compact (find, describe, run)](https://ai-powered.com/mcp/agent-data-tools-compact-find-describe-run): Agent Data Tools: Compact (find, describe, run) is a freemium MCP server for developer tools.
- [Angel Agent Guard: pre-purchase check + prompt-injection scan](https://ai-powered.com/mcp/angel-agent-guard-pre-purchase-check-prompt-injection-scan): Angel Agent Guard: pre-purchase check + prompt-injection scan is a freemium MCP server for ai models and llm tooling.
- [Mizan Calibration](https://ai-powered.com/mcp/mizan-calibration): Mizan Calibration is a free MCP server for ai models and llm tooling.

---
Source: https://ai-powered.com/mcp/xf-model-source
