Nous Research, the New York-based AI collective known for its “personalized, unrestricted” language models, has launched a new Inference API to make its models more accessible to developers and researchers through a programmatic interface.
This API launch marks a major expansion of Nous Research’s capabilities, challenging the more restricted AI approaches of industry giants like OpenAI and Anthropic. The company shared on social media: “We heard your feedback and built a simple system to make our language models more accessible to developers and researchers everywhere.
The initial API release features two flagship models:
- Hermes 3 Llama 70B – A general-purpose AI model based on Meta’s Llama 3.1 architecture.
- DeepHermes-3 8B Preview – A reasoning model that enables users to toggle between standard responses and detailed chains-of-thought (CoT).
To manage demand, Nous has implemented a waitlist system, with access granted on a first-come, first-served basis. New users receive $5 in free credits, and API documentation is available for developers looking to integrate the service into their applications.
Unlike AI giants with vast GPU resources, Nous Research operates with infrastructure limitations common to smaller AI labs. The waitlist approach serves both as a technical necessity and a marketing tactic, creating an exclusive appeal while ensuring efficient computational resource management.
Despite branding itself as an alternative to big tech AI, Nous is also adopting pragmatic business strategies to scale inference services sustainably. This balancing act between idealism and practicality will likely shape its evolution from an open-source AI hub to a commercial AI provider.
Nous Research’s API follows a structure similar to OpenAI’s API for completions and chat completions, making it easier for developers already familiar with OpenAI’s interface to integrate Nous’ models into their applications.
From Open-Source Models to Cloud API: Nous Research’s Business Shift
This API launch comes just four months after Nous introduced Nous Chat, its first user-facing chatbot interface. While Nous has previously released numerous open-source models for local deployment, this new API allows developers to leverage high-performance versions without the complexity of self-hosting.
Additionally, Nous’ latest DeepHermes-3 model, released last month, focuses on reasoning-based AI, allowing users to switch between concise responses and detailed explanations via system prompts.
Challenging AI Guardrails: Nous Research’s Unrestricted Philosophy
Since its founding in 2023, Nous Research has positioned itself as an alternative to more tightly controlled AI systems. The company advocates for greater user autonomy and transparency, as reflected in its blog posts like “Freedom at the Frontier” and “From Black Box to Glass House: The Imperative for Transparent AI Development.”
Despite marketing itself as “unrestricted AI”, Nous Research still incorporates some safety mechanisms to prevent harmful outputs, striking a balance between freedom and responsible AI deployment.
Monetizing Open AI Research: The Roadmap for Hermes and DeepHermes
The API launch signifies Nous Research’s transition toward a sustainable business model, while remaining committed to open-source AI development. Since July 2023, Nous has released 29 AI artifacts, including models, research papers, code, and datasets.
This strategy aligns with Nous’ hybrid business model, where:
- Individual developers and researchers can still download and run models locally.
- Enterprises seeking scalability and optimization can pay for API access.
By monetizing infrastructure and optimization rather than restricting access to model weights, Nous is attempting to generate revenue while preserving its open-source ethos.
Nous Research’s API strategy raises an important question: Can independent AI labs establish sustainable business models without relying on big tech or venture capital pressures?
As Nous expands its inference offerings, future integrations could include:
- Hermes 2 Pro – A model specializing in function-calling.
- Psyche Project – A decentralized AI research initiative on Solana.
By competing with established AI providers like Together AI, Anthropic, and OpenAI, Nous’ new API could drive further innovation in the AI inference space.