DeepSeek V4 Pro Discount Explained
DeepSeek's official pricing page lists DeepSeek V4 Pro with a temporary 75% discount extended until 2026-05-31 15:59 UTC.
That makes DeepSeek V4 Pro one of the most important current price checks for cost-sensitive teams.
Discounted prices
| Model | Cache-hit input / 1M | Cache-miss input / 1M | Output / 1M | Context |
|---|---|---|---|---|
| DeepSeek V4 Flash | $0.0028 | $0.14 | $0.28 | 1M |
| DeepSeek V4 Pro | $0.003625 | $0.435 | $0.87 | 1M |
DeepSeek also lists the non-discounted V4 Pro prices as 1.74 cache-miss input, and $3.48 output per 1M tokens.
Why cache hits matter
The gap between cache-hit and cache-miss input price is large. A repeated system prompt, retrieval scaffold, or agent instruction block can change the bill materially if it is billed as cache-hit input.
Use the DeepSeek V4 Pro token calculator with your real input/output mix.
Compatibility note
DeepSeek says deepseek-chat and deepseek-reasoner will be deprecated in the future and correspond to non-thinking and thinking modes of deepseek-v4-flash. New pricing pages should point users toward:
- DeepSeek V4 Flash token calculator
- DeepSeek V4 Pro token calculator
- DeepSeek V4 Pro vs GPT-5.4 mini pricing
Practical rule
If the 75% discount is still active for your billing window, V4 Pro deserves a fresh cost check. If you are planning beyond 2026-05-31, model both the discounted and non-discounted prices before making a long-term migration decision.