What is an LLM?

LLM definition (simple)

An LLM (Large Language Model) is a machine-learning model trained on a huge corpus of text. It learns to predict the next token — roughly, the next word or sub-word — and in doing so it learns to answer questions, write, summarize, translate, and help with code.

Examples you may know: GPT-4, Claude, Gemini, Llama, Qwen, Mistral. Most people interact with them through cloud chat apps. LlamaBox lets you run compatible open-weight models directly on an Android phone.

From cloud LLM to local LLM

A local LLM is the same kind of model, but the file is stored on your device and inference happens on your hardware. LlamaBox loads a quantized GGUF file into app memory and runs it with llama.cpp on the phone CPU. Nothing is sent to a chat API.

Key terms

• Parameters — the "size" of a model, often 0.5B, 1B, 3B, 7B, etc. More parameters usually means more capability but more RAM and slower inference.
• Quantization — compressing weights to fewer bits. Q4_K_M is a popular mobile format that keeps most quality while cutting size roughly in half versus float16.
• GGUF — the file format used by llama.cpp to store quantized model weights and metadata.
• Context window — how many tokens the model can "remember" at once. LlamaBox defaults to 2048.

What can a local LLM on Android do?

• Draft notes, messages, and journal entries offline
• Answer questions without sending data to a vendor
• Summarize text you paste into the app
• Help with coding questions in offline environments
• Analyze images with a vision-capable model + mmproj

Tradeoffs vs cloud chatbots

Next: GGUF models for Android · how to run an LLM on Android · free LLMs you can run locally · offline AI Android.

FAQ