Millie by LLMs For All vs Local LLM runner

Millie is a private AI assistant that runs entirely on your phone. There is no server behind the chat. No account, no sign-up. Once you’ve downloaded a model, Millie works without an internet connection—on a plane or anywhere without a signal. Ask Millie to help you think something through: how to word a message you’re dreading sending, what to cook with what’s in the fridge, a gift for someone hard to buy for, a wedding toast, or an email you’ve been putting off. Choose between Millie, our flagship 35-billion-parameter language model built for deeper understanding and thoughtful answers, and Millie Flash, our lighter model built for the fastest replies. The flagship model can also understand photos and offers adjustable reasoning effort. Personalization lets you tell Millie about yourself and how you’d like it to respond. Because Millie runs on your phone instead of in a data center, it’s smaller than cloud AI. It’s weaker on niche facts and recent events, and it can be wrong, so check anything that matters. What it’s good at is thinking things through with you, privately. Before you install: The app requires at least 6 GB of RAM. That covers all iPhone 14 models and later, plus the Pro models of iPhone 12 and 13. iPhone 11 and earlier, the standard iPhone 12 and 13, and all iPhone SE models are unsupported. Millie Flash runs on supported phones with 6 GB. The flagship model requires at least 8 GB. The app shows which models your phone supports before downloading. The initial download is about 2 GB for Millie Flash or 8.6 GB for the flagship model. Download time depends on your connection, and chatting becomes available once the download finishes. Millie is made by LLMs For All, Inc.

Run LLM on your iPhone, 100% local, no network (Network needed once to download a model) Models: • Qwen2.5 0.5B Instruct — ~0.4 GB (4 GB RAM) • Llama 3.2 1B Instruct — ~0.8 GB (4 GB RAM) • Qwen2.5 1.5B Instruct — ~0.9 GB (4 GB RAM) • Llama 3.2 3B Instruct — ~1.9 GB (6 GB RAM) • Qwen3 4B Instruct 2507 — ~2.5 GB (6 GB RAM) • Qwen2.5 7B Instruct — ~4.4 GB (8 GB RAM) Also: • Add your own GGUF (for example, from Hugging Face) • Auto-estimated max context from GPU memory; you can change it Requires iOS 18 or later. Tested on iPhone 11, 14 Pro, 15 Pro with iOS 26. Qwen models © Alibaba Cloud. Licensed under Apache License 2.0. Llama 3.2 is licensed under the Llama 3.2 Community License, Copyright © Meta Platforms, Inc. All Rights Reserved. Built with Llama. On-device inference uses llama.cpp (MIT License).