# Vetted Consumer

Vetted Consumer is an MCP server for checking whether local language models fit your hardware, choosing GGUF quantization, and comparing GPU purchase and rental costs.

- MCP server · `type=mcp` · Claude Code, Codex, Cursor, GitHub Copilot, Windsurf (Devin Desktop), Cline, Zed, Gemini CLI · 2408 installs
- Author: hi-10f9
- License: not stated · policy=link-only · description only; the file stays with the author
- Source: https://smithery.ai/server/hi-10f9/vetted-consumer
- Updated: 2026-07-17
- Security check: automated, 2026-11-16 (text) — no signs of malicious code found.
- HTML: https://affpapa.org/en/nevibecode/mcp/vetted-consumer

## What it does

The server runs as a hosted HTTP endpoint without requiring local package installations or API credentials. It connects your assistant directly to a maintained catalog containing specs for Macs, desktop GPUs, and mini-PCs alongside model architectures.

Built-in functions calculate whether a weights file fits your available memory and context window. Tools like `recommend_quant` display token rates and file sizes across compression levels, while `cost_compare` calculates break-even periods and electricity costs. Pricing data for used graphic cards refreshes monthly from marketplace listings.

Because the tool relies on a hosted service, requests will fail if there is no outbound network connection.

## Requirements
- Remote server: `https://vetted-consumer--hi-10f9.run.tools`

## Install

### Claude Code (native platform)

**Platform CLI: remote server**
```bash
claude mcp add --transport http vetted-consumer https://vetted-consumer--hi-10f9.run.tools
```

### Codex

**Platform CLI: remote server**
```bash
codex mcp add vetted-consumer --url https://vetted-consumer--hi-10f9.run.tools
```

### Cursor

**Config file**
File `.cursor/mcp.json`:
```json
{
    "mcpServers": {
        "vetted-consumer": {
            "url": "https://vetted-consumer--hi-10f9.run.tools"
        }
    }
}
```

### GitHub Copilot

**Config file**
File `.vscode/mcp.json`:
```json
{
    "servers": {
        "vetted-consumer": {
            "type": "http",
            "url": "https://vetted-consumer--hi-10f9.run.tools"
        }
    }
}
```

### Gemini CLI

**Platform CLI: remote server**
```bash
gemini mcp add -t http vetted-consumer https://vetted-consumer--hi-10f9.run.tools
```

## When to use it
- determine if a local model runs on your current GPU or Mac
- select the optimal GGUF quantization tier for your memory limit
- calculate break-even timelines between buying hardware and paying API fees
- find the most affordable machine capable of running a model at full context

## Example prompts
- Will Mixtral 8x7B fit on my RTX 4070 Ti, and how many tokens per second will it produce?
- What GGUF quantization should I download for Command R on a 32 GB unified memory machine?
- Compare 1-year costs of buying a used GPU versus running an API model.

## Similar
- [serverless-deploy](https://affpapa.org/en/nevibecode/skills/serverless-deploy.md) — skill · 3700 installs
- [VRAMGlass](https://affpapa.org/en/nevibecode/mcp/vramglass.md) — MCP server · ★ 1 repo stars
- [hf-mem](https://affpapa.org/en/nevibecode/skills/hf-mem.md) — skill · 1000 installs
- [model-quantization](https://affpapa.org/en/nevibecode/skills/model-quantization.md) — skill · 199 installs
- [runpod-vllm-claude-code](https://affpapa.org/en/nevibecode/skills/runpod-vllm-claude-code.md) — skill · 1400 installs
- [hardware-probe](https://affpapa.org/en/nevibecode/mcp/hardware-probe.md) — MCP server · ★ 7 repo stars

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md
