Introduction

llamadart is a Dart and Flutter plugin for running llama.cpp models with GGUF files across native and web targets.

Who this is for#

• App developers building local-first AI features in Dart/Flutter.
• Teams that need OpenAI-style HTTP compatibility from local models.
• Maintainers who need predictable native/web runtime integration.

Core primitives#

• LlamaEngine: stateless generation API.
• ChatSession: stateful chat wrapper over LlamaEngine.
• LlamaBackend: platform backend abstraction used by the engine.

Read by workflow#

• First setup: Installation
• First inference: Quickstart
• Multi-turn chat: First Chat Session
• Function calling: Tool Calling
• Template diagnostics: Chat Templates and Parsing
• Template internals: Template Engine Internals
• LoRA runtime workflows: LoRA Adapters
• Performance work: Performance Tuning
• Platform/backend planning: Platform & Backend Matrix
• Upgrade planning: Upgrade Checklist
• Maintainer operations: Maintainer Overview