DeepSeek Off-Peak
official pricing ↗

Current billing window

Local Beijing · billing

Rates USD per 1M tokens

Model Input · cache hit Input · cache miss Output

Off-peak is 50% of peak — peak is 2× off-peak. Cache-hit inputs are ~31× cheaper than cache-miss — reuse your prompts. ¥ rate indicative (fixed 7.1).

Cost calculator

Off-peak$0
Peak$0
You save$0
Automate off-peak dispatch