35% below DeepSeek's published list price.

Where the number comes from

Our rate is 35% below DeepSeek's published list price. Their price list is public — open it and check the arithmetic yourself. We would rather you did. What follows is DeepSeek API pricing as it applies to your workload: the published list is where DeepSeek pricing starts, not where your own number lands.

What we cannot put on a page is your figure. Off-peak scheduling bills batch work at half the peak rate, and prompt caching cuts the cost of repeated input sharply. How much of your workload can use either one is what moves the number — and that differs by customer. So we ask two questions: how many tokens a month, and what are you running today.

What we can tell you up front

These are properties of the model's published pricing, not of our quote — they apply whatever rate we agree.

• Off-peak calls are billed at half the peak rate
• Cached input tokens are billed far below fresh ones
• You pay per token — no seat licenses, no platform fee
• One monthly invoice, payable by bank transfer
• Prices exclude applicable taxes