When using the DeepSeek-R1 model on cloud platforms like AWS, Microsoft Azure, or Google Cloud, there are several factors to consider regarding additional fees, particularly for data transfer.
1. AWS: When deploying models like DeepSeek-R1 using AWS Bedrock Custom Model Import, the primary costs are based on Custom Model Units (CMUs) for inference and storage. However, AWS typically charges for data transfer out of its network. For example, data transfer out of AWS to the internet is charged at a rate that varies by region, typically starting at $0.09 per GB for the first 10 TB per month[1]. There is no specific mention of additional data transfer fees directly related to the DeepSeek-R1 model itself, but general AWS data transfer costs apply.
2. Microsoft Azure: Azure does not provide specific pricing for DeepSeek-R1 in its standard documentation, and it recommends contacting the sales or support team for detailed pricing information[6]. However, Azure generally charges for data transfer out of its network, similar to AWS. The cost depends on the region and the amount of data transferred.
3. Google Cloud: Like AWS and Azure, Google Cloud charges for data transfer out of its network. The pricing varies by region and the amount of data transferred. For instance, data transfer out to the internet is typically charged at a rate starting at $0.12 per GB for the first 10 TB per month. There is no specific mention of additional data transfer fees for DeepSeek-R1, but standard Google Cloud data transfer costs apply.
4. DeepSeek API: When using the DeepSeek-R1 model via its parent company's API, there are no specific additional fees mentioned for data transfer. The pricing is primarily based on input and output tokens, with costs such as $2.19 per million output tokens[2][3].
In summary, while there are no specific additional fees for data transfer directly related to the DeepSeek-R1 model itself, users should consider the general data transfer costs associated with each cloud platform. These costs can add up, especially for large-scale deployments or when transferring significant amounts of data out of the cloud networks.
Citations:
[1] https://repost.aws/questions/QU-hcixrtFSaSoKH8GL-KogA/pricing-model-of-deepseek-r1-distilled-llama-models-with-amazon-bedrock-custom-model-import
[2] https://campustechnology.com/Articles/2025/02/04/AWS-Microsoft-Google-Others-Make-DeepSeek-R1-AI-Model-Available-on-Their-Platforms.aspx?admgarea=news
[3] https://prompt.16x.engineer/blog/deepseek-r1-cost-pricing-speed
[4] https://www.reddit.com/r/LocalLLaMA/comments/1id2jia/psa_deepseekr1_is_available_on_nebius_with_good/
[5] https://livechatai.com/deepseek-r1-pricing-calculator
[6] https://learn.microsoft.com/en-us/answers/questions/2152620/price-for-usage-of-deepseek-r1
[7] https://www.linkedin.com/pulse/running-full-deepseek-r1-model-home-cloud-stephan-janssen-2cpyf
[8] https://api-docs.deepseek.com/quick_start/pricing