> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.nvidia.com/nemoclaw/latest/user-guide/hermes/inference/local-inference/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.nvidia.com/nemoclaw/_mcp/server.

# Local Inference

## Docs

- [Choose a Local Inference Server](https://docs.nvidia.com/nemoclaw/latest/user-guide/hermes/inference/local-inference/choose-local-inference-server.md): Compare operator-run servers and managed runtime profiles before choosing local inference for NemoClaw.
- [Set Up Ollama](https://docs.nvidia.com/nemoclaw/latest/user-guide/hermes/inference/local-inference/set-up-ollama.md): Install, connect, and configure Ollama as NemoClaw's local inference server.
- [Set Up vLLM](https://docs.nvidia.com/nemoclaw/latest/user-guide/hermes/inference/local-inference/set-up-vllm.md): Connect NemoClaw to an existing vLLM server or start a managed vLLM container.
- [Set Up vLLM on Two DGX Sparks](https://docs.nvidia.com/nemoclaw/latest/user-guide/hermes/inference/local-inference/set-up-vllm-on-two-dgx-sparks.md): Qualify two trusted DGX Spark systems and start NemoClaw's Experimental distributed DeepSeek V4 Flash vLLM profile.
- [Set Up vLLM on Two DGX Stations](https://docs.nvidia.com/nemoclaw/latest/user-guide/hermes/inference/local-inference/set-up-vllm-on-two-dgx-stations.md): Qualify a trusted two-Station DGX GB300 pair and use NemoClaw's distributed Nemotron Ultra vLLM recipe.
- [Set Up llama.cpp](https://docs.nvidia.com/nemoclaw/latest/user-guide/hermes/inference/local-inference/set-up-llama-cpp.md): Attach an existing llama.cpp server or select an experimental managed llama.cpp profile.
- [Set Up NVIDIA NIM](https://docs.nvidia.com/nemoclaw/latest/user-guide/hermes/inference/local-inference/set-up-nvidia-nim.md): Pull, start, and configure a local NVIDIA NIM container for NemoClaw inference.