AWS becomes the first cloud service provider to deliver the fully managed DeepSeek-R1 LLM model as generally available within its Amazon Bedrock offering.
Amazon Web Services (AWS) has announced that the fully managed serverless DeepSeek-R1 AI model, which disrupted the AI and tech market in January, is now generally available on Amazon Bedrock – a cloud computing platform that simplifies the building of generative AI apps.
DeepSeek-R1 models first became available on Amazon Bedrock on January 30, through its Marketplace and Custom Model Import. Since then, thousands of users who are managing their own infrastructure have started using these models. Customers appreciate the strong security measures and useful tools that ensure safe AI deployment, so they often pair the new tool with features like Amazon Bedrock Guardrails, to boost security.
Now, Amazon is making DeepSeek even easier to use by offering more options, including a new serverless solution. This allows businesses of all sizes to use this advanced AI tool without needing technical expertise. “With this fully managed implementation, customers can utilize the model in a serverless pay-per-token fashion that helps them scale from experiment to production without managing any infrastructure,” said Vasi Philomin, VP of generative AI, AWS.
Besides, AWS customers can access not only DeepSeek-R1 but also its smaller, distilled versions (models designed to mimic DeepSeek-R1) in Amazon Bedrock. Additionally, businesses can upload their fine-tuned DeepSeek-R1 Distilled Llama models and run them as fully managed solutions using Amazon Bedrock Custom Model Import, which allows them to integrate customized AI models through a single API. DeepSeek-R1 current availability options are making it easier for businesses of all sizes to use the technology without worrying about setup or maintenance.
DeepSeek-R1 is one of the AI models developed by the Chinese startup DeepSeek. It is publicly available and designed for advanced reasoning, offering high accuracy and a strong understanding of context when processing complex tasks. According to the AWS website, DeepSeek-R1, with its strong reasoning skills, has shown top performance in mathematics, scoring 79.28% on AIME 2024, and in software engineering, scoring 49.2% on SWE-bench Verified. This powerful model gives businesses an affordable way to use advanced, scalable AI solutions that can easily adjust to their specific needs.
Customers can use DeepSeek-R1 for solving complex problems, coding, data analysis, and more while benefiting from enterprise-grade security, including data encryption, access controls, and compliance safeguards. AWS also provides Amazon Bedrock Guardrails solution, as well as Amazon Bedrock model evaluation features, to supplement the LLM and help prevent inaccurate AI-generated content.
Amazon’s safeguards also include filtering sensitive information and adjustable security settings, especially useful for organizations in highly regulated industries. Amazon Bedrock Guardrails safeguards can be customized to the business’s application requirements and responsible AI policies. This helps block harmful language, explicit content, or anything inappropriate for users, as well as identify and remove personal data to keep user privacy safe.
