You can now try DeepSeek R1 on Amazon Bedrock

Amazon Web Service is quickly picking up on the AI trend as Amazon BedRock, its AI platform, is adopting the DeepSeek R1 model.

AWS announced that developers can now deploy the distilled versions of DeepSeek-R1 language models via Bedrock. The distillation process makes larger models like DeepSeek-R1 more efficient by reducing computational requirements while maintaining close performance levels.

You can also have access to both proprietary and open models.

“Smaller distilled models like the 8B version can process requests much faster and consume fewer resources, making them more cost-effective for production deployments,” Amazon explains.

“Whereas larger distilled versions like the 70B model maintain closer performance to the original while still offering meaningful efficiency gains,” the company says further.

To deploy DeepSeek-R1-Distill-Llama models on Amazon Bedrock, you need an AWS account, IAM permissions, and an S3 bucket for storing the model. First, download the model files from Hugging Face and upload them to your S3 bucket.

Then, you’d need to import the model directly into Amazon Bedrock using the S3 path and create an import job. After importing, test the model in the Amazon Bedrock Playground or via APIs.

And they’re not the only American tech firm adopting Chinese technology. Microsoft has previously announced that it’s integrating NPU-optimized DeepSeek R1 models into its Copilot+ PCs, starting with Qualcomm’s Snapdragon X chips and later expanding to Intel Core Ultra 200V.

These models will run locally on devices using the Windows Copilot Runtime and AI Toolkit. They’re also available for developers to try via Azure AI & GitHub.