# Overview

Rent a GPU on Clore.ai, run a local LLM via Ollama or vLLM, and connect a coding assistant — you get a fully **private AI development environment** where your code never leaves the machine. No API keys to manage, no token limits, no data sent to third-party servers.

## How It Works

```
Clore.ai GPU  →  Ollama / vLLM (local LLM)  →  Coding Tool (Aider, TabbyML)
                     ↑                              ↑
              RTX 4090 / A100               connects via localhost:11434 or :8000
```

1. **Rent a GPU** on [clore.ai/marketplace](https://clore.ai/marketplace) — RTX 3090 ($0.30–1/day) or RTX 4090 ($0.50–2/day)
2. **Deploy an LLM** — `ollama run deepseek-r1:32b` or spin up vLLM with any coding-oriented model
3. **Launch your coding tool** — it talks to the LLM over localhost, completing code, writing tests, and refactoring

## Available Guides

| Guide                                                         | Tool  | Description                                                                                |
| ------------------------------------------------------------- | ----- | ------------------------------------------------------------------------------------------ |
| [Aider](https://docs.clore.ai/guides/ai-coding-tools/aider)   | Aider | Terminal-based AI pair programmer — edits files in your repo directly via natural language |
| [TabbyML](https://docs.clore.ai/guides/ai-coding-tools/tabby) | Tabby | Self-hosted code completion server with IDE extensions (VS Code, JetBrains)                |

## Why Self-Host on Clore.ai?

* **Privacy** — your codebase stays on the rented instance, not on OpenAI/Anthropic servers
* **No rate limits** — unlimited completions at GPU speed
* **Cost control** — pay by the hour/day, spin down when idle
* **Model choice** — run any open model: DeepSeek-R1, Qwen 2.5 Coder, CodeLlama, StarCoder2