DeepSeek API Pricing Guide (2026)
Estimate DeepSeek Flash costs with cached-input assumptions and a peak-time usage mix.
Verified November 9, 2026. Direct DeepSeek API, USD.
Flash pricing
| API ID | Off-peak input | Off-peak cache read | Off-peak output | Peak multiplier |
|---|---|---|---|---|
| deepseek-flash | 0.15 | 0.003 | 0.6 | 2× across all three categories |
Official price ↗. The API ID deepseek-flash serves V4.1-Flash. The official rate card separately lists V4-Pro; it is not included in this calculator’s current selection.
Peak and off-peak estimates
Peak windows are 01:00–04:00 and 06:00–10:00 UTC, Monday–Friday, excluding Chinese public holidays. Other hours are off-peak.
For the same input/output mix and caching ratio in both periods: blended cost = off-peak cost × (1 + peak token share). A 50% peak share gives 1.5× the off-peak estimate. It does not mean 50% of clock hours or 50% of requests unless request sizes are equal.
Monthly example
| Requests | All off-peak | 50% peak | All peak |
|---|---|---|---|
| 10,000 | $4.5 | $6.75 | $9 |
| 100,000 | $45 | $67.5 | $90 |
| 1,000,000 | $450 | $675 | $900 |
Examples assume 1,000 input and 500 output tokens per request, no cache hits. Cache writes, tools, taxes and other services are excluded.
Choosing a model
A lower token price does not establish equal quality. Test accuracy, latency, failure rates and integration compatibility on your own tasks.