DeepSeek V4 Pro API pricing — what it really costs
DeepSeek V4 Pro (DeepSeek) costs $1.32 per 1M input tokens and $3.96 per 1M output tokens — with prompt caching, cached input drops to $0.044. Peak-hour price (01-04 and 06-10 UTC, Mon-Fri); off-peak costs half. Prices verified 2026-10-17; we track the price you are actually billed today, including promotional rates.
Price per 1M tokens
| Input | Output | Cached input |
|---|---|---|
| $1.32 | $3.96 | $0.044 |
Real monthly costs on DeepSeek V4 Pro
Three realistic workloads, computed from the prices above (cached share billed at the cache-read rate where the provider offers one):
Customer-support chatbot
$3.09/mo
2M input + 0.5M output tokens/month, 60% of input served from prompt cache
RAG / document Q&A
$12.50/mo
20M input + 1M output tokens/month, 70% cached (shared context)
Agentic coding assistant
$34.76/mo
50M input + 5M output tokens/month, 80% cached (long sessions)
Want exact numbers for your usage? Open the calculator — same verified prices, your token volumes.
Compare DeepSeek V4 Pro
- DeepSeek V4 Pro vs Claude Haiku 4.5 — which is cheaper?
- DeepSeek V4 Pro vs GPT-5.4 Mini — which is cheaper?
- DeepSeek V4 Pro vs DeepSeek V4.1 Flash — which is cheaper?
- DeepSeek V4 Pro vs Claude Sonnet 5 — which is cheaper?
FAQ
How much does the DeepSeek V4 Pro API cost?
DeepSeek V4 Pro costs $1.32 per 1M input tokens and $3.96 per 1M output tokens. Cached input tokens cost $0.044 per 1M. Note: Peak-hour price (01-04 and 06-10 UTC, Mon-Fri); off-peak costs half. Pricing verified 2026-10-17.
What does a typical workload cost on DeepSeek V4 Pro?
Customer-support chatbot: about $3.09/month (2M input + 0.5M output tokens/month, 60% of input served from prompt cache). RAG / document Q&A: about $12.50/month (20M input + 1M output tokens/month, 70% cached (shared context)). Agentic coding assistant: about $34.76/month (50M input + 5M output tokens/month, 80% cached (long sessions))
Is DeepSeek V4 Pro expensive compared to other models?
DeepSeek V4 Pro ranks #8 of 23 models by blended price in our comparison. The cheapest tracked models are Claude Haiku 5.5 and GPT-6 Luna at $0.100/$0.500 per 1M tokens.