Features
Run multiple open-source models locally (Mistral, Llama, Phi, etc.)
Up to 5 concurrent model sessions (multi-weave)
Contextual thread fusion to chain model outputs
Universal model manager supporting Hugging Face, Ollama, GGUF, GPTQ
Automatic quantization optimization (4-bit, 8-bit, FP16)
Native file looming indexing up to 100K tokens
Real-time collaborative editing over local WebSocket
Intelligent prompt looms for multi-model code review
Cross-platform support: Windows (x64), macOS (Intel+Apple Silicon), Linux (x64/ARM)
Multilingual interface in 12 languages
24/7 offline availability with no server dependencies
Privacy-first offline mesh with no telemetry or cloud relays
No user accounts or subscription required
Supports fast weft (<200ms response) with 1-3B models
Supports warp thread with 7-70B models for complex tasks
@ai_fn decorator for LLM-powered Python functions
@ai_classifier decorator for text classification
Structured data extraction via Pydantic models
Agent loops with tool calling and function calling
Built-in streaming (SSE) support
Async-first API for concurrent applications
Rate limiting and automatic retries
LLM response caching for performance
Concurrency control for workload management
CLI tool for monitoring and debugging
SQLite state store for persistence
OpenAI and Anthropic model support
Embeddings generation for semantic search
Local execution with no cloud dependency
Self-hosted deployable as a library
