openclaw2go lets you run your OpenClaw agent in a self-contained environment powered by local open source models. Pick your models, see exactly how much VRAM you need, and deploy with a single Docker command.

Features

Multi-Model Support — Run LLMs, image models, and audio models side by side
VRAM Calculator — See real-time memory usage across your selected models
Hardware Detection — Supports NVIDIA GPUs (4090, 5090, L40, A100, H100, B200) and Apple Silicon (M3 Pro, M4, M4 Pro, M4 Max, M4 Ultra)
One-Click Deploy — Get a Docker run command ready to go, or deploy via MLX
Cross-Platform — Works on Linux, Windows, and macOS
Model Browser — Search and filter hundreds of models by task type, quantization, context length, and memory

How It Works

Select your operating system and available VRAM
Browse and pick models for each modality (LLM, image, audio)
Review VRAM usage and hardware compatibility
Copy the Docker command and deploy

Stack

llama.cpp for LLM inference
diffusers for image generation
Docker for containerized deployment