China AI Company DeepSeek Making a move again! At… OpenAI、Anthropic In an environment where major players’ API pricing remains high and computational costs continue to rise, DeepSeek has taken the opposite approach. Whether to celebrate the launch of its new flagship V4 series just days ago or not, it recently announced permanent price cuts across its entire API lineup, slashing input (cache hit) prices for all models, including the V4 series, to just one-tenth of their original rates. More importantly, for the flagship V4 Pro, there’s an additional limited-time discount of 75% off. For those using AI agents, this is a great opportunity to experience DeepSeek V4 Pro during the promotional period.
DeepSeek API Announces Permanent Price Cuts! Only 1/10 of the Original Price, V4 Pro Adds a Limited-Time Extra 75% Off Discount
This DeepSeek API promotion is divided into two parts: “DeepSeek’s entire model series input (cache hit) permanently reduced to 1/10 the price” and “V4 Pro additional 75% off limited-time offer”
What I mean by DeepSeek’s entire model lineup is that not only V4 Pro and V4 Flash, but also older models like V3.x have all been permanently reduced in price. Also, some people might not understand what “input (cache hit)” means.
DeepSeek API’s pricing structure is essentially divided into three categories: “Input (Cache Hit)”, “Input (Cache Miss)”, and “Output”. A “Cache Hit” means your prompt’s prefix matches content the system has recently processed, allowing you to apply the lower Cache Hit rate. This entire determination is automatically handled by DeepSeek’s backend, so users don’t need to configure anything manually. A “Cache Miss”, of course, means the content being processed is completely different.
So, using DeepSeek V4 as an example, below is the difference before and after the permanent price reduction for input (cache hit) (V4 Pro doesn’t have the additional 75% discount applied yet):
- DeepSeek-V4-Pro input (cache hit) from 1 RMB → 0.1 RMB
- DeepSeek-V4-Flash input (cache hit) from 0.2 RMB → 0.02 RMB
And DeepSeek-V4-Pro’s extra 2.5-fold limited-time discount means all three items are “2.5-fold off,” so after conversion it becomes:
- Input (Cache Hit): 0.1 yuan RMB → 0.025 yuan RMB
- Input (cache miss): from 12 CNY → 3 CNY
- Output: From 24 yuan RMB → 6 yuan RMB
The discount is quite substantial. The offer ends on May 5 at 23:59 (Beijing Time), lasting a total of 9 days.
Image source: ITHOME
V4 Pro was already significantly cheaper than its competitors GPT-5.5 and Claude Opus 4.7, and now with the permanent price cut plus a 75% discount, the price gap has widened by more than just one tier—you practically don’t have to worry about costs at all, so go ahead and use it as much as you want.
If you want to learn more about the features of DeepSeek V4, you can refer to our previous report:
Source: KOCPC Chinese