NemoClaw

Self-hosted AI inference environment combining OpenClaw gateway with NVIDIA Nemotron-3 Super 120B model.

Description

Custom setup for running AI inference locally/on own infrastructure. OpenClaw provides the gateway layer, NVIDIA Nemotron provides the model.

Download or use

Deployed on own infrastructure via Docker/k3s.

Reasoning for

Independence from cloud AI providers for specific workloads. Used in Agentic Systems architecture.

Key components:

  • OpenClaw — gateway and sandbox environment (Docker k3s)
  • NVIDIA Nemotron-3 Super 120B — large language model for inference
  • Cloudflare Tunnel — remote access

Alternatives considered

  • Ollama — simpler but less capable models
  • Cloud-only (Claude API, OpenAI) — easier but vendor-dependent

Resources

NVIDIA Nemotron


Template: tool