Hermes Unlocks Self-Improving AI Agents, Powered by NVIDIA RTX PCs and DGX Spark

Hermes Agent, developed by Nous Research, is an open source agentic AI framework that has gained 140,000 GitHub stars in under three months and is now the most-used agent on OpenRouter. It stands out for self-evolving skills (writing and refining its own capabilities over time), contained sub-agents for cleaner task isolation, curated reliability, and consistently stronger results than competing frameworks on identical models. Hermes is optimized for local, always-on use on NVIDIA RTX PCs, RTX PRO workstations, and DGX Spark. The post also highlights Qwen 3.6 27B and 35B models, which match or exceed much larger predecessor models while fitting in far less memory, making them ideal for local agent workloads. Setup is straightforward via llama.cpp, LM Studio, or Ollama.