We are excited to announce the arrival of the Meta Llama 3 8B Instruct model on Private LLM, a local chatbot app available now for iOS devices with 6GB or more of RAM and macOS. On iOS, we offer a 3-bit quantized version, while on macOS, we provide a 4-bit quantized model. The macOS version works on any Intel or Apple Silicon Mac. This new AI model by Meta is available on Pro and Pro Max iOS devices as recent as iPhone 13 Pro, including full 8K context length on the iPhone 15 Pro with 8GB of RAM.

Llama 3 brings a significant reduction in false refusal rates, allowing for broader discussions with fewer interruptions. This chatbot engages in more open dialogues and supports a wider variety of prompts. Whether you're digging into niche topics or everyday queries, Llama 3 handles your conversations better.

Continuing our commitment to privacy, Llama 3 operates entirely offline. This ensures that your data remains secure within your device, never leaving it for any reason. The local nature of this chatbot sets it apart from cloud-based alternatives.

From our tests, we noticed that Llama 3 is good at writing code as well. This makes it an ideal companion for coding projects, whether you are working on building apps, scripting, or automating tasks. The GPT-based model excels at understanding and generating code.

Screenshot of Meta Llama 3 8B Instruct Running Locally on iPhone — Meta Llama 3 8B Instruct Running Locally on iPhone

Screenshot of Meta Llama 3 8B Instruct Running on Mac — Meta Llama 3 8B Instruct Running on Mac

Screenshot of Meta Llama 3 8B Instruct Running on iPad — Meta Llama 3 8B Instruct Running on iPad

If you're looking for an uncensored Meta Llama 3 8B fine-tune, we've introduced Uncensored Dolphin 2.9 Llama 3 8B Model on Private LLM. This model lets you have unrestricted, uncensored, and even NSFW conversations.

Private LLM also offers several fine-tuned versions Llama 3 8B model, such as Llama 3 Smaug 8B, Llama 3 8B based OpenBioLLM-8B, and Hermes 2 Pro - Llama-3 8B, on both iOS and macOS.

For Apple Silicon Macs with more than 48GB of RAM, we offer the bigger Meta Llama 3 70B model. With this model, users can experience performance that rivals GPT-4, all while maintaining privacy and security on their devices.

Download Meta Llama 3 8B Instruct on iPhone, iPad, or Mac:

Get the latest version of Private LLM app from the App Store.
Go to Settings > Models and Choose 'Llama 3 8B Instruct' to download it onto your device.

Please note that Meta Llama 3 requires a Pro/Pro Max iPhone, an iPad with M-series Apple Silicon, or any Intel or Apple Silicon Mac.