DeepSeek
deepseek-v4-pro API Pricing
DeepSeek charges $0.66 per 1M input tokens and $1.98 per 1M output tokens across a 1M context window. Verified 2026-08-17 against the official DeepSeek pricing page.
Standard off-peak rate (cache-miss input); cache-hit input significantly discounted; peak hours 01:00-04:00 and 06:00-10:00 UTC double pricing
Input / 1M tokens
$0.66
Output / 1M tokens
$1.98
Context window
1M
What it costs to run
| Workload | Tokens | Cost |
|---|---|---|
| Single short call | 2K in, 500 out | $0.00231 |
| Typical chat turn | 10K in, 1K out | $0.00858 |
| Long document pass | 100K in, 2K out | $0.070 |
| 1,000 chat turns | 10M in, 1M out | $8.58 |
Price history
Input price has moved +52% since tracking began on 2026-06-11.
| Effective | Input / 1M | Output / 1M |
|---|---|---|
| 2026-08-17 | $0.66 | $1.98 |
| 2026-06-11 | $0.43 | $0.87 |
How it compares
Ranked on the blended cost of a typical chat turn (10K in, 1K out), not the input rate alone.
| gemini-3.5-flash-lite | $0.00550 per turn | |
| OpenAI | gpt-4.1-mini | $0.00560 per turn |
| OpenAI | gpt-3.5-turbo | $0.00650 per turn |
| OpenAI | gpt-3.5-turbo-0125 | $0.00650 per turn |
| gemini-3-flash-preview | $0.00800 per turn | |
| DeepSeek | deepseek-v4-pro | $0.00858 per turn |
| gemini-3.7-flash | $0.011 per turn | |
| gemini-3.6-flash | $0.011 per turn | |
| gemini-3.8-flash | $0.011 per turn | |
| gemini-3.1-flash-live-preview | $0.012 per turn | |
| OpenAI | gpt-3.5-turbo-1106 | $0.012 per turn |
Head to head
Full cost breakdowns against the models deepseek-v4-pro is usually weighed against, including the workload shape where the cheaper option flips.
TechFastForward coverage of deepseek-v4-pro
What we wrote when this model, or the rest of its line, was news.
- DeepSeek V4 Pro Undercuts US AI Coding Cost 95% 20262026-06-05
- Qwen3.7 Max Surpasses Gemini Flash on AI Index 20262026-06-05
- DeepSeek V4 Pro Doesn't Just Match Claude, It Does It at 86% Off the Sticker Price2026-05-09
- DeepSeek V4 on Huawei Chips Is Not a Benchmark Story. It Is a Supply Chain Story That Changes Everything.2026-05-07
- NIST Says DeepSeek Is 8 Months Behind. DeepSeek Disagrees. And Both Are Right in Ways That Should Worry You.2026-05-04
- DeepSeek V4 Cuts Frontier AI Coding Cost 75 Percent2026-06-02
Free JSON API
Every tracked model, including this one, is available as JSON. Free to use with attribution and a link back.
curl https://techfastforward.com/api/data/llm-pricingMaintained by Jordan Hale, TechFastForward. Prices are read daily from the official provider page and recorded only when they move. See every tracked model on the LLM API Pricing Tracker.