# Nodegrove VRAM: can I run it?

> Nodegrove VRAM: can I run it? is a free MCP server for ai models and llm tooling. Check whether an open LLM fits your GPU, with VRAM math and what fits instead. It can run locally on your machine or as a hosted remote server.

- **Kind:** MCP server
- **Pricing:** Free
- **Platforms:** API, CLI
- **Website:** https://nodegrove.io/mcp
- **Repository:** https://github.com/nodegrove/vram-mcp
- **MCP Registry name:** io.nodegrove/vram-mcp
- **Categories:** AI models and LLM tooling
- **Listing updated:** 2026-11-14

## About

Every figure is a stated formula over a model's config.json and a card's published specs, with no benchmarks or guesses. A sample answer says a 4090 cannot run Llama 3.3 70B at Q4_K_M, shows the 45.8 GB needed against 22.8 GB available, then suggests a larger card or a smaller model that does fit.

The data covers 40 models and 40 GPUs, and any other Hugging Face repo can be read from its config.json. Names are matched the way people type them, such as 4090 or M4 Max, and an unlisted card can be described by its memory. Answers include the memory split, a speed ceiling, the longest context that fits and, on a no, every change that would make it fit. Estimates are labelled as estimates.

It is free, read-only and needs no account or key, available as a remote endpoint or a local stdio package needing Node.js 20.

## Features

- **Fit verdicts:** Says fits, tight or no, with the memory split.
- **What-fits search:** Checks every model on one card and recommends one.
- **VRAM estimates:** Weights, KV cache and overhead per quantisation.
- **Hugging Face repos:** Reads any repo's config.json for memory needs.
- **Fix suggestions:** Lists changes that would make a no into a yes.

## Good for

- Check a GPU before downloading a large model
- Find the best model for a 16 GB card
- Estimate memory for a long context window
- Compare cards for a planned local setup

## Install

### Claude Code (Terminal)

```sh
claude mcp add --transport http vram-mcp https://mcp.nodegrove.io/mcp
```

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

```sh
https://mcp.nodegrove.io/mcp
```

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

```json
{
  "mcpServers": {
    "vram-mcp": {
      "url": "https://mcp.nodegrove.io/mcp"
    }
  }
}
```

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

```json
{
  "servers": {
    "vram-mcp": {
      "type": "http",
      "url": "https://mcp.nodegrove.io/mcp"
    }
  }
}
```

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

```json
{
  "mcpServers": {
    "vram-mcp": {
      "serverUrl": "https://mcp.nodegrove.io/mcp"
    }
  }
}
```

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

```json
{
  "mcpServers": {
    "vram-mcp": {
      "type": "http",
      "url": "https://mcp.nodegrove.io/mcp"
    }
  }
}
```

## Tools

- `can_i_run`: Says whether a model fits a card, with the memory split, speed ceiling and context limit.
- `what_fits`: Checks every model on one card and gives recommendations.
- `estimate_vram`: Weights, KV cache and overhead at each quantisation.
- `estimate_from_hf_repo`: Memory estimate for any Hugging Face repo from its config.json.
- `list_models`: Lists known models with specs.
- `list_gpus`: Lists known GPUs with specs.

## FAQ

### Does it need an account?

No account or key; it is free and read-only.

### How many models and cards are covered?

40 models and 40 GPUs, plus any Hugging Face repo.

### Can I run it locally?

Yes, over stdio with npx and Node.js 20 or newer.

### Are the numbers benchmarks?

No, they are stated formulas and are labelled estimates.

### Is Nodegrove VRAM: can I run it? free?

Yes. Nodegrove VRAM: can I run it? is free to use.

### How do I add Nodegrove VRAM: can I run it? to Claude Code?

Run this command in your terminal: claude mcp add --transport http vram-mcp https://mcp.nodegrove.io/mcp

### What are the best alternatives to Nodegrove VRAM: can I run it??

The closest alternatives to Nodegrove VRAM: can I run it? 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/nodegrove-vram-can-i-run-it
