Features
One-command self-hosted setup (./install.sh)
Isolated Docker containers per sandbox
Per-sandbox filesystem and memory limits
Live preview URLs with automatic routing and TLS
Sandbox idle sleep and instant wake
Pre-installed OpenCode and Claude Code CLIs
REST API for create sandbox, exec, write files
Streaming results over SSE
SQLite-based state management
Traefik HTTPS routing and termination
Reconciler converges Docker state on reboot
Host-memory pressure reaper for multi-tenant safety
Custom API provider keys per sandbox
MIT licensed for commercial use
Fast & lossless inference for 1-bit LLMs (BitNet b1.58)
Optimized CPU kernels for ARM & x86 architectures
Official GPU inference kernel (released 05/2025)
Parallel kernel implementations with configurable tiling
Embedding quantization for 1.15x–2.1x additional speedup
1.37x–6.17x CPU speedup vs baseline
55%–82% CPU energy reduction
Run 100B BitNet b1.58 on single CPU (5-7 tok/s)
Lookup Table kernels built on T-MAC methodologies
Support for Hugging Face 1-bit models
Conda environment setup script (setup_env.py)
Inference server (run_inference_server.py)
Lossless inference—no accuracy degradation
Falcon3 family and Llama3-8B-1.58 model support
Supports models up to 100B parameters
