How to Run DeepSeek R1 & Llama 3.3 Locally with Ollama and Open-WebUI

# How to Run DeepSeek R1 & Llama 3.3 Locally with Ollama Running high-performance reasoning models locally ensures 100% data privacy, zero API rate limits, and offline availability. --- ## 1. Hardware Requirements & Quantization | Model Size | Quantization | Minimum RAM / VRAM | Recommended Hardware | |---|---|---|---| | 8B / 7B | Q4_K_M | 8 GB | Apple M1/M2/M3 (16GB RAM) or RTX 3060 | | 14B / 32B | Q4_K_M | 24 GB | Apple M-series (36GB+ RAM) or RTX 4090 | | 70B | Q4_K_M | 48 GB | Mac Studio M2 Ultra (64GB+) or 2x RTX 3090 | --- ## 2. 3-Minute Setup with Ollama ```bash # 1. Install Ollama via Homebrew brew install ollama # 2. Start the local server ollama serve # 3. Pull and run DeepSeek R1 distilled model ollama run deepseek-r1:14b ``` --- ## 3. Connecting to IDE & Web Interfaces You can integrate your local Ollama server directly into **Cursor**, **Continue.dev**, or deploy **Open-WebUI** using Docker for a self-hosted ChatGPT interface.