Dolphin AI - Local Setup Guide

⚠️ Notice

This guide is solely meant for educational purposes. It is your responsibility how you use this information—use AI sensibly.

Why Use Dolphin AI?

Dolphin Uncensored is a modified AI model that removes standard content restrictions, allowing for unfiltered responses.
For those not comfortable with a local setup, Venice AI (Premium Plan required) offers a cloud-based alternative.

System Requirements

Before starting, ensure your system meets the requirements.

Minimum Specifications:

• CPU: AMD Ryzen 5 or Intel i5 (6th Gen or newer)
• RAM: 16GB DDR4
• GPU: NVIDIA GTX 1660 or RTX 2060 (6GB VRAM)
• Storage: SSD with at least 50GB of free space
• Operating System: Linux, macOS, or Windows 10/11

Recommended Specifications:

• CPU: AMD Ryzen 9 or Intel i7 (10th Gen or newer)
• RAM: 32GB+ DDR4/DDR5
• GPU: AMD 7900XTX (24GB+ VRAM) or NVIDIA RTX 3090 / 4090
• Storage: NVMe SSD with 100GB+ free space
• Operating System: Latest Linux distros, macOS, or Windows 11

Model Variants and Hardware Demands

Dolphin AI supports multiple model sizes, each with its own hardware needs:

• Dolphin-Llama3:8B → Requires 5GB VRAM

  dolphin-llama3 ollama run 3:8b
  

• Dolphin-Llama3:70B → Requires 40GB VRAM

  dolphin-llama3 ollama run 3:70b
  

• Notice No GPU? You can use CPU instead, but expect slower response times

  Installation and Setup

• Ollama is required to manage local language models

  Download and install from: (https://ollama.com)

• verify:

  ollama --version
  

  Download Dolphin AI

  ollama pull dolphin-llama3
  
  

  you can run the model by :

  ollama run dolphin-llama3:8b   # For 8B model7
  

  ollama run dolphin-llama3:70b  # For 70B model
  

  for a user-friendly interface instead of terminal Chatbox AI

• Download: (https://chatboxai.app/en#download)

Install and open the application and Go to Settings > Local LLM, and link it to Ollama

• You have total control over your AI experience when you run Dolphin AI locally. Self-hosting gives you control over whether you want to use it for learning, research and more hahaha…..