AWS, Google Cloud, and Microsoft Launch DeepSeek R1 via Managed Services

Businesses can access DeepSeek R1 through managed services provided by AWS, Google Cloud and Microsoft as they deliver secure deployment of advanced AI models.

The AWS Bedrock service now incorporates DeepSeek's complete R1 model which enables customers to work with this robust AI model. The announcement by AWS builds upon its earlier release of miniature AI models Llama and Qwen for the customer base. AWS CEO Matt Garman stated that various models remain unfit for distinct usage scenarios as numerous new models are expected to develop.

The DeepSeek R1 model is accessible for experimentation through Google Cloud Vertex managed service and Microsoft has included the model within Audio Foundry and GitHub. These models can be accessed by major cloud service providers because organizations increasingly need customizable AI solutions.

DeepSeek-R1 needs at least 800 GB of HBM memory for its inference process and users can deploy it through SageMaker JumpStart and the Bedrock Marketplace provided by AWS. To secure their models Bedrock users need to integrate ApplyGuardrail API before using it because AWS explains such security requirements are crucial for companies who worry about outside model security within their region.

Hugging Face enables deployment of DeepSeek R1 through its Inference Endpoints which provides businesses with secure affordable infrastructure to run machine learning models on AWS platforms.

The newly released AI managed services bring businesses access to advanced AI models which ensure flexibility and scalability and deliver security features across their AI implementations.