Blog

Notes from on-device land.

Field-notes on shipping GGUF models on Android and iOS with llamafu.

• Mar 11, 2026 structured-output · tool-calling

  Structured output on-device with grammars and schemas

  How llamafu uses GBNF grammars and JSON schemas to get reliable structured output from quantized models running on a phone.

• Feb 4, 2026 gguf · performance

  Picking a GGUF quantization for mobile

  Q4_K_M, Q4_0, Q8_0 — what they mean, when to pick which, and how llamafu lets you load any of them.

• Jan 14, 2026 on-device · mobile

  Why on-device LLMs are no longer a science project

  Four practical reasons to move LLM inference onto the phone, and what llamafu gives you to do it on Flutter.