n8n runs fully local AI workflows on a single NVIDIA DGX Spark

n8n published a deployment guide for running AI workflows entirely on one NVIDIA DGX Spark without cloud AI APIs. The setup combines local models (Qwen3.6-35B-A3B-NVFP4, Whisper, OCR, embeddings), LiteLLM request routing and an OpenAI-compatible endpoint for n8n orchestration.

• The guide was published on October 19, 2026
• A LiteLLM proxy blocks requests to cloud providers
• Stack includes Qwen3.6-35B-A3B-NVFP4, Whisper, OCR and embeddings
• NVentures joined n8n's Series C in October 2025