bridge library
The low-level iOS/macOS plugin bridge, backed by llama.cpp.
All native work runs through a Pigeon bridge that is driven from a dedicated worker isolate, so model loading and token streaming never block the UI.
Most consumers want the neutral, cross-platform API in
package:llama_cpp_flutter/llama_cpp_flutter.dart instead; this entrypoint
is the raw bridge it wraps, kept separate so Apple-specific surface stays
out of the cross-platform API.
Classes
- LlamaBridgeSession
- A loaded model session, returned by LlamaCppFlutter.loadModel.
- LlamaCppFlutter
- Entry point for llama.cpp inference.
- LlamaGenerationStats
- Token accounting and timing for one completed generation run.
Enums
- LlamaFinishReason
- Why a generation run stopped.
Exceptions / Errors
- LlamaException
- Thrown when native model loading or generation fails.