Skip to main content

DeepSeek's Holiday Gift: Ten Days of Half-Price API Calls

DeepSeek's Holiday Gift: Ten Days of Half-Price API Calls

If you're a developer who relies on DeepSeek's API, your wallet is about to get a break. On September 19, the company cleared up a nagging question about its peak and off-peak pricing: weekends that double as adjusted workdays—and all national holidays—are billed at the cheaper off-peak rate. That means from September 20 through October 10, you'll enjoy a ten-day stretch of half-price calls, the longest such window this year.

What exactly counts as off-peak?

DeepSeek's pricing scheme already made weekends cheap. But adjusted workdays—like Sunday, September 20, or Saturday, October 10—sat in a gray zone. The new clarification removes all doubt: any weekend, even if it's a make-up workday, gets off-peak pricing. The same goes for the Mid-Autumn Festival (September 25–27) and National Day (October 1–7). Add those two adjusted weekend days, and you've got more than ten consecutive days of discounted rates.

During peak hours on regular weekdays—9:00 to 12:00 and 14:00 to 18:00—you pay double. Off-peak, that cost is cut in half. So if you've been putting off a big project or testing a new feature, now's the time to fire up those API calls.

Image

A new model that sweetens the deal

As if the timing weren't good enough, DeepSeek released its V4.1 Flash model on September 10, and it's a game-changer for cost-conscious developers. This 552-billion-parameter MoE model uses an asymmetric architecture—activating just 8 billion parameters for input and 16 billion for output. Despite its smaller active footprint, it beats the flagship V4Pro on several benchmarks. And the price? During off-peak hours, input cache hits cost just 0.02 yuan per million tokens, with output at 4 yuan per million tokens.

That's not just cheap—it's practically a steal for the performance you get.

Why this matters for your bottom line

For startups and indie developers, API costs can make or break a project. A ten-day half-price window combined with a model that punches above its weight class means you can experiment more, scale faster, and worry less about the bill. Whether you're building a chatbot, a code assistant, or something entirely new, this is the moment to push your ideas forward.

So mark your calendar: September 20 to October 10. Set a reminder. And maybe treat yourself to that extra test run you've been putting off. Your future self—and your budget—will thank you.

Key Points:

  • DeepSeek confirms adjusted workdays and national holidays are billed at off-peak rates.
  • From September 20 to October 10, developers get over ten days of half-price API calls.
  • Off-peak pricing is half the cost of peak hours (weekdays 9-12, 14-18).
  • The new V4.1 Flash model offers top-tier performance at rock-bottom prices: 0.02 yuan per million input tokens (cache hit) and 4 yuan per million output tokens during off-peak.
  • This is the most concentrated period of discounted rates this year—ideal for testing and scaling AI projects.