DeepSeek's token pricing structure is designed to be competitive and cost-effective, particularly when compared to other AI models like OpenAI's offerings. Here are the key details:
Pricing Overview
- Currency Options: DeepSeek provides pricing in both USD and CNY.- Cost Calculation: Pricing is based on the number of tokens used, specifically calculated per million tokens.
Models and Pricing
1. DeepSeek Chat Model:- Input Price: $0.14 per million tokens (for cache misses, $0.014 for cache hits).
- Output Price: $2.20 per million tokens.
- Context Length: 64K tokens.
- Max Output Tokens: 8K.
2. DeepSeek Reasoner Model:
- Input Price: $0.55 per million tokens (for cache misses, $0.14 for cache hits).
- Output Price: $2.19 per million tokens.
- Context Length: 64K tokens.
- Max Output Tokens: 8K with 32K max CoT tokens.
Cost Comparisons
- DeepSeek's pricing is significantly lower than that of OpenAI's models, with DeepSeek being about 27 times cheaper for input tokens and nearly 27.4 times cheaper for output tokens compared to OpenAI's GPT-4 model[1][2].- For example, generating a response using 10 million tokens would cost approximately $200 with DeepSeek compared to $900 with ChatGPT, highlighting a substantial savings of around $700 per month for high-volume users[3][4].
Additional Features
- Context Caching: This feature optimizes costs by reducing prices during repeated queries (cache hits) versus new queries (cache misses) which incur higher costs[1].- Discounts: There are promotional discounts available until February 8, 2025, though these do not apply to the DeepSeek R1 model[1].
In conclusion, DeepSeek's pricing structure is tailored for affordability and efficiency, especially beneficial for developers and businesses looking to manage AI-related costs effectively while leveraging advanced AI capabilities.
Citations:
[1] https://meetcody.ai/blog/deepseek-r1-api-pricing/
[2] https://www.vellum.ai/blog/the-training-of-deepseek-r1-and-ways-to-use-it
[3] https://www.creolestudios.com/deepseek-vs-chatgpt-cost-comparison/
[4] https://play.ht/blog/deepseek-vs-chatgpt/
[5] https://play.ht/blog/deepseek-pricing/
[6] https://www.interconnects.ai/p/deepseek-v3-and-the-actual-cost-of
[7] https://artificialanalysis.ai/models/deepseek-v3
[8] https://daily.dev/blog/deepseek-everything-you-need-to-know-about-this-new-llm-in-one-place