DeepSeek just made it even cheaper for developers to use its AI model

Table of Contents
Ever since the Chinese AI startup DeepSeek released its R1 model last month, it has remained in the spotlight, continuing to make big moves that have shaken up the industry. A recent example of this is its collaboration with Nvidia, where the joint venture aims to optimize DeepSeek R1 for Nvidia's Blackwell architecture, potentially boosting revenue by 25x while lowering costs by 20x.
And if there weren't already headlines about DeepSeek's lower costs, the startup has now introduced discounted off-peak pricing for developers looking to use its AI models to build their own products. Previously, domestic rivals like Alibaba have countered DeepSeek by lowering prices for their AI services, and now this latest move from the Chinese startup could once again put significant pressure on competitors both in China and overseas.
API usage costs to be up to 75% cheaper
The company announced on Wednesday that between 16:30 GMT and 00:30 GMT, the cost of using its API would be up to 75% cheaper. For context, DeepSeek currently operates in two forms: one as a chatbot service and enterprise API directly managed by DeepSeek, and the other as an open-source ‘DeepSeek-R1' model that anyone can download to integrate into apps and web products.
Today's best deals
- Intel Core Ultra 245K - 15% OFF NOW!
- ASUS ROG Swift PG32UQXR - $200 OFF NOW!
- Yeyian Yumi RTX 4060 Gaming PC - $500 OFF NOW!
- SAMSUNG 990 PRO 4TB SSD - 35% OFF NOW!
- Sony X77L 4K Smart TV - 16% OFF NOW!
- Samsung Galaxy Tab A9+ - 29% OFF NOW!
- WD_BLACK 8TB SN850X SSD - 32% OFF NOW!
*Stock availability and pricing subject to change depending on retailer or outlet.
According to DeepSeek, API usage costs during this timeframe will be reduced by 75% for the R1 model and 50% for the V3 model. This means the V3 model, which normally costs up to $1.10 per million output tokens, will now be priced at $0.55, while the R1 model, usually $2.19 per million output tokens, is now just $0.55. While the company refers to this period as “off-peak” since it runs from 00:30 to 08:30 in Beijing, it actually covers daytime hours in Europe and the United States.
DeepSeek keeps making its services more accessible
With that said, DeepSeek’s open approach has been a highlight for a while now, showing how you can train with limited data on a consumer GPU like the RX 7900 XTX without spending millions. With this new “off-peak” pricing, DeepSeek seems to make its services not only accessible to the general public but also to developers who want to bring the power of AI into their products.
This wave of accessibility has also pushed other major AI companies like OpenAI into high gear, as they recently transformed their AI chatbot ChatGPT into an AI-powered search engine. So, while the fight for AI dominance continues to get more competitive, consumers and developers are sure to see more advantages come out of it.