DeepSeek makes the V4 Pro price discount permanent (api-docs.deepseek.com)

Article summary

DeepSeek has made the price discount for their V4 Pro model permanent. The model's pricing is based on the number of input and output tokens, with a preference for using granted balance first when both balances are available. The company reserves the right to adjust prices, and users are recommended to top up based on their actual usage. The model's details and pricing information are available on the DeepSeek website.

Main themes

• DeepSeek V4 Pro pricing
• Model usage and costs
• Token-based pricing
• Data retention and privacy
• Model performance and comparison

What commenters say

• DeepSeek's V4 Pro model offers better performance than other models for complex coding tasks at a lower cost.
• The model's token-based pricing can lead to significant cost savings, especially with the permanent price discount.
• Some users are concerned about data retention and privacy when using DeepSeek's models, and are looking for providers with no-data-retention policies.
• The use of third-party gateways or providers may affect the pricing and performance of DeepSeek's models.
• DeepSeek's caching mechanism can significantly reduce costs and improve performance, especially for agent workflows.
• The company's decision to make the price discount permanent is seen as a positive move, making their model more competitive in the market.
• Some users believe that using a model with a coding agent from the same company can provide better performance and integration.
• Others argue that using off-the-shelf coding agents can work just as well with DeepSeek's models, and that the company's own coding agent may not be necessary.