Nous Research just launched an API that gives developers access to AI models that OpenAI and Anthropic won’t build

Nous Research, a prominent AI collective based in New York known for its groundbreaking work in developing personalized and unrestricted language models, has recently launched a new Inference API. This API aims to make Nous Research’s models more accessible to developers and researchers through a programmatic interface.

The API release marks a significant expansion of Nous Research’s offerings, challenging the more restricted approaches adopted by larger AI companies such as OpenAI and Anthropic. The company announced on social media that they have built a simple system to address user feedback and enhance accessibility to their language models.

The initial release of the API features two flagship models from Nous Research – Hermes 3 Llama 70B and DeepHermes-3 8B Preview. Hermes 3 Llama 70B is a powerful general-purpose model based on Meta’s Llama 3.1 architecture, while DeepHermes-3 8B Preview is a reasoning model that allows users to switch between standard responses and detailed chains-of-thought.

To manage the high demand for their services, Nous Research has implemented a waitlist system through their new portal. Access to the API is granted on a first-come, first-serve basis, with new accounts receiving $5 in free credits. Developers can refer to the API documentation on the portal to explore integration options.

The waitlist approach not only addresses technical constraints faced by smaller AI organizations like Nous Research but also serves as a marketing tactic to create exclusivity and manage computational load. This strategic positioning reflects Nous Research’s grassroots ethos and pragmatic business strategies as they transition from open-source releases to commercial offerings.

The API launch follows the design pattern of OpenAI’s API for completions and chat completions, making it easier for developers familiar with that interface to integrate Nous Research’s models into their applications.

This API launch comes on the heels of Nous Research’s debut of Nous Chat, their first user-facing chatbot interface. While the company has released several open-source models for local deployment, the new API enables developers to access high-performance versions of these models without the need to manage their own infrastructure.

Since its establishment in 2023, Nous Research has positioned itself as an alternative to tightly controlled AI systems, emphasizing individual agency and alignment with user needs. The company’s ‘unrestricted AI’ philosophy challenges big tech’s guardrails while incorporating guardrails against harmful outputs.

The API launch signifies Nous Research’s move towards a sustainable business model while upholding open-source principles. By commercializing deployment and continuing to release model weights, Nous aims to generate revenue without compromising its core principles. This hybrid approach caters to individual developers and researchers as well as enterprises seeking reliability and convenience.

Nous Research indicates that their inference offerings will expand over time, potentially including more models like Hermes 2 Pro and their Psyche project. The new API provides AI startups with an alternative to established players like Together AI, Anthropic, and OpenAI, fostering competition and driving innovation in the AI inference space.

In conclusion, Nous Research’s API launch signifies a step towards a more sustainable business model while maintaining their commitment to open-source values. The company’s community-oriented approach to AI development invites collaboration and innovation to shape the future of AI technology.