Home
Distributed orchestration platform for local language models.
Synapse Engine coordinates multiple local LLMs into a unified inference system. Instead of running models one at a time, it orchestrates them across performance tiers-fast models for initial processing, powerful models for refinement-with sub-100ms contextual retrieval built in.
It runs entirely on your hardware. No API keys, no cloud dependencies, no data leaving your machine.
Key capabilities:
- Multi-model orchestration across FAST/BALANCED/POWERFUL tiers
- Query modes: Simple, Two-Stage, Council (Consensus/Debate), Benchmark
- CGRAG: Sub-100ms contextual retrieval with FAISS
- Metal acceleration for Apple Silicon
- WebUI-first design - full control in the browser
- Docker Quick Start - Full installation and setup
- Docker Quick Reference - Command cheat sheet
- Model Management UI - Using the web interface
- Docker Quick Start - Installation and configuration
- Docker Quick Reference - Common commands
- Admin Panel Guide - System administration
- Troubleshooting - Common issues and solutions
- Model Management UI - Discovering and managing models
- Profile System - Configuring model profiles
- Dynamic Model Control - Start/stop models without restart
-
Query Modes Overview - All available query processing modes:
- Simple Mode - Single model, direct response
- Two-Stage Mode - Fast tier + CGRAG retrieval, then refinement
- Council Mode - Multi-model collaboration or debate
- Benchmark Mode - Compare models side-by-side
- Dynamic Model Control - Live model management without Docker restarts
- WebTUI Style Guide - UI styling standards
- WebTUI Integration Guide - Terminal UI implementation
| Query Modes | Model Management |
|---|---|
 |
 |
| System Status | Architecture |
|---|---|
|
 |