Recently, AWS Marketplace (China Region) officially launched the DeepSeek series model APIs. With an on-demand, pay-as-you-go pricing model, this offering aims to reduce the computational costs of AI application development and large-scale deployment, allowing enterprises to focus more on product innovation and market expansion.
AWS official website: https://www.amazonaws.cn/
DeepSeek Model Development Timeline
Since the release of DeepSeek-V3 in December 2024, the DeepSeek team has continuously improved AI inference capabilities. On January 20, 2025, DeepSeek introduced DeepSeek-R1, along with a series of DeepSeek-R1-Distill models ranging from 1.5B to 70B parameters.
Currently, DeepSeek-R1 is available on Amazon Bedrock and Amazon SageMaker AI. Users can also directly access these models via the AWS Marketplace (China Region), operated by NWCD (Ningxia Western Cloud Data).
At the AWS re:Invent 2024 conference, CEO Andy Jassy stated that Amazon has internally developed over 1,000 generative AI applications, emphasizing that “no single model can fit all use cases.” As a result, AWS continues to expand its AI model ecosystem, incorporating both established industry models and emerging high-performance solutions—DeepSeek being a key part of this strategy.
DeepSeek Model Advantages
1. Industry-Leading Inference Engine
DeepSeek leverages a self-developed high-efficiency architecture and optimized framework, significantly improving inference throughput while reducing latency, making it ideal for high-concurrency and low-latency enterprise scenarios.
2. Flexible Scalability for Multiple Use Cases
DeepSeek supports dynamic scaling, allowing users to flexibly adjust workloads and deploy customized models with ease to meet diverse business requirements.
3. End-to-End Optimization to Reduce AI Costs
Through full-stack optimization, DeepSeek reduces both inference and deployment costs. Its tiered pricing model helps enterprises avoid resource waste and maintain precise budget control.
4. Enterprise-Grade High Availability
DeepSeek features comprehensive monitoring systems and fault-tolerant mechanisms, along with professional technical support, ensuring high availability for enterprise users.
5. Intelligent Operations and Cost Analysis
With built-in auto-scaling and cost analysis capabilities, DeepSeek dynamically adjusts computing resources based on workload demands, helping businesses optimize AI deployment and improve efficiency.
6. Security and Compliance for Enterprise Needs
DeepSeek adopts multi-layered security measures, including compute isolation, network isolation, and storage isolation, meeting industry standards and ensuring enterprise data security.
DeepSeek Pricing Details
| Model Name | Input (per 1M Tokens) | Output (per 1M Tokens) | Context Length |
|---|---|---|---|
| DeepSeek-R1 (Pro) | ¥4 | ¥16 | 64K |
| DeepSeek-V3 (Pro) | ¥2 | ¥8 | 64K |
| DeepSeek-R1-Distill-Qwen-1.5B (Pro) | ¥0.14 | ¥0.14 | 32K |
| DeepSeek-R1-Distill-Qwen-7B (Pro) | ¥0.35 | ¥0.35 | 32K |
| DeepSeek-R1-Distill-Llama-8B (Pro) | ¥0.42 | ¥0.42 | 32K |
| DeepSeek-R1-Distill-Qwen-14B | ¥0.70 | ¥0.70 | 32K |
| DeepSeek-R1-Distill-Qwen-32B | ¥1.26 | ¥1.26 | 32K |
| DeepSeek-R1-Distill-Llama-70B | ¥4.13 | ¥4.13 | 32K |
Note: Actual pricing is subject to the console display.
Free Trial of 40+ AWS Services
AWS offers developers free access to over 40 enterprise-grade cloud services, helping businesses rapidly build AI and cloud applications.
1. EC2 Cloud Servers – 12-Month Free Tier
- Up to 750 hours per month
- Two instance types available
- Includes 750 hours of IPv4 public address usage
2. S3 Object Storage – 12-Month Free Tier
- 5GB standard storage
- 20,000 GET requests and 2,000 PUT requests free
3. AWS Lambda – Always Free
- 1 million free requests per month
- 3.2 million seconds of compute time
4. API Gateway – 12-Month Free Tier
- 1 million API calls per month free
👉 Click here to visit the AWS official website and start your free trial: https://aws.amazon.com/cn/campaigns/nc20241201
Frequently Asked Questions (FAQ)
Q1: What are the main differences between DeepSeek-R1 and DeepSeek-V3?
DeepSeek-R1 features a more optimized inference architecture, offering higher throughput and lower latency, making it suitable for real-time inference scenarios.
DeepSeek-V3 is better suited for general AI tasks, such as text generation and code completion.
Q2: How is DeepSeek pricing calculated?
DeepSeek uses a token-based pricing model, charging based on input and output tokens consumed during API calls. Refer to the pricing table above for details.
Q3: What industries can benefit from DeepSeek models?
DeepSeek models are widely applicable across industries, including customer service, code generation, market analysis, medical text processing, and financial risk control, providing powerful AI capabilities.
Q4: How can I get started with DeepSeek models?
Enterprises can register via the AWS Marketplace (China Region) and enable the DeepSeek API, then quickly integrate it through Amazon Bedrock or Amazon SageMaker AI.
Summary
The launch of DeepSeek-R1 (Pro) and DeepSeek-V3 (Pro) by AWS provides enterprises with a more efficient, flexible, and cost-effective AI deployment solution. Whether for large-scale inference workloads or optimizing AI application costs, businesses and developers can leverage DeepSeek to build more intelligent AI services.
