DeepSeek R1 Expands to Nvidia, AWS, and GitHub, Revolutionizing AI Access

DeepSeek has made significant progress in expanding the accessibility of its flagship model, DeepSeek R1.

After gaining considerable attention in the AI community, the model is now available on major platforms including Nvidia, AWS, and GitHub.

This move marks a milestone for DeepSeek, which has already captured the interest of developers with its open-source approach.

Broad Integration Across Major Platforms

DeepSeek R1 is now integrated into several top tech platforms, making it accessible to a wide range of developers. Nvidia has incorporated the model as a NIM microservice, leveraging its Hopper architecture for real-time, high-quality responses.

The integration utilizes FP8 Transformer Engine acceleration and NVLink connectivity, enabling DeepSeek R1 to generate up to 3,872 tokens per second on an HGX H200 system.

AWS offers DeepSeek R1 through Amazon Bedrock, simplifying API integration. Additionally, Amazon SageMaker provides the opportunity for more advanced customization and training.

With the support of AWS Trainium and Inferentia, the deployment ensures cost efficiency. AWS also offers DeepSeek-R1-Distill, a lighter version of the model, which simplifies infrastructure management while maintaining scalability.

Collaborative Model Development and Future Plans

On GitHub, DeepSeek has expanded its presence, with over 3,300 DeepSeek-based models available on the collaborative AI-model development platform Hugging Face.

This allows developers to build and refine models based on DeepSeek’s architecture, fostering further innovation in the AI space.

Microsoft has also taken significant steps in supporting DeepSeek, incorporating it into its Azure AI Foundry. Microsoft’s Azure platform provides developers with a secure and scalable environment to integrate AI into their workflows.

The company is also working on implementing distilled versions of DeepSeek R1 for local deployment on Copilot+ PCs in the future, further expanding its reach.

Cost-Effective and Powerful AI Model

One of the most significant aspects of DeepSeek R1 is its cost-efficiency. Despite its powerful capabilities, including 671 billion parameters and a 128,000-token context length, DeepSeek R1 was trained for just $6 million.

This makes it significantly cheaper to train compared to similar models from industry giants like NVIDIA and Microsoft, making it a game-changer for AI development.

DeepSeek’s advanced reasoning capabilities have already garnered praise and positioned it as a strong competitor to popular AI models like ChatGPT.