Nous Research
Nous Research is a leading innovator in human-centric AI systems, specializing in advanced simulators, open-source language models, and next-generation reasoning frameworks. Founded in 2023 as an AI accelerator and research lab, the company focuses on building AI that closely reflects real-world human behavior by emphasizing cutting-edge model architecture, synthetic data generation, fine-tuning techniques, and multi-step reasoning. Nous Research champions an open, transparent approach to artificial intelligence, prioritizing the development of open-source, human-compatible LLMs that stand in contrast to today’s dominant closed-model ecosystems. The company is best known for the Hermes series, a collection of general-purpose, human-aligned, lightweight language models that have been downloaded more than 50 million times on Hugging Face and widely adopted across the AI research community. Beyond Hermes, Nous has introduced several breakthrough technologies and infrastructure tools, including DisTrO, Forge API, Nous Chat, and Nous Pysche.
ABOUT NOUS RESEARCH
Nous Research is a leading innovator in open-source AI, decentralized training infrastructure, and human-aligned model development.
The company’s mission is to democratize access to advanced artificial intelligence by building models that are transparent, community-driven, and trained through distributed compute rather than traditional centralized data centers.
Leveraging blockchain technology—specifically the Solana network—Nous Research coordinates and incentivizes global contributors to supply GPU resources for large-scale training.
This decentralized approach positions Nous as a pioneer in community-powered AI, distributed LLM training, and next-generation open-source AI ecosystems.
NOTABLE PRODUCTS & SERVICES
Psyche Network – A decentralized AI training network built on the Solana blockchain. Psyche allows individuals and organizations to contribute idle GPU capacity to collaboratively train large AI models, creating a crowdsourced, permissionless compute layer for scalable model development.
DeepHermes-3 (LLM) – An advanced open-source large language model featuring toggleable reasoning modes. Users can switch between an intuition-based fast mode and a slower, more analytical reasoning mode, enabling customizable performance for coding, analysis, research, and creative tasks.
Forge Reasoning API – A next-generation reasoning framework combining Monte Carlo Tree Search (MCTS), Chain-of-Code (CoC) reasoning, and a Mixture of Agents (MoA) architecture. Forge enables flexible, multi-step, and creative problem-solving across a variety of applications.
TEEHEE Autonomous Agent Experiment – An exploratory project designing an autonomous AI agent with a persistent digital identity operating inside a Trusted Execution Environment (TEE). The experiment explores secure, verifiable, and ethically aligned agentic behavior.
Atropos Coordination System – A distributed coordination and scheduling system that manages inference tasks across large numbers of worker nodes. According to Nous Research, Atropos enables efficient orchestration of large-scale generation workloads in decentralized environments.
Behavioral Simulation Platform – Nous Research conducts long-horizon behavioral simulations such as WorldSim, Doomscroll, and Gods & S8n to study how AI agents develop socially, ethically, and strategically in controlled role-playing environments. These simulations offer insights into AI safety, agent alignment, and emergent behaviors.