DeepSeek API Pricing: Costs per Million Tokens
DeepSeek charges for API input and output tokens. Your cost depends on the model, cache hits and the time of use. Compare the official rates and see worked examples below.
Official DeepSeek API prices
USD per 1 million tokens. Input cache hits, input cache misses and output are billed separately.
Checked 2026-11-10
| Model | Period | Input cache hit | Input cache miss | Output |
|---|---|---|---|---|
| DeepSeek V4.1 Flashdeepseek-flash | Off-peak | $0.003 | $0.15 | $0.6 |
| DeepSeek V4.1 Flashdeepseek-flash | Peak | $0.006 | $0.3 | $1.2 |
| DeepSeek V4 Prodeepseek-v4-pro | Off-peak | $0.022 | $0.66 | $1.98 |
| DeepSeek V4 Prodeepseek-v4-pro | Peak | $0.044 | $1.32 | $3.96 |
Rates may change. Check the official Models & Pricing page before making a budget. DeepSeek V4 Hub is independent and is not the official DeepSeek service.
How DeepSeek API billing works
Input, output and cache
Input is the content sent to the model; output is what the model generates. Reused input served from cache is charged at the cache-hit rate, while the remainder uses the cache-miss rate. A cache hit is not guaranteed.
Peak and off-peak
Peak hours are 09:00–12:00 and 14:00–18:00 Beijing time (01:00–04:00 and 06:00–10:00 UTC), Monday–Friday, excluding Chinese public holidays. All other hours, weekends, and Chinese public holidays are off-peak. Official off-peak rates are half the peak rates.
Cost formula: cached input × cache-hit rate + uncached input × cache-miss rate + output × output rate. Divide each token count by one million before applying the rates.
DeepSeek Search and Fetch are separate APIs: each is listed at $0.75 per 1,000 successful requests. Their charges are additional to model usage; see the official pricing and billing rules.
DeepSeek API cost examples
These examples use hypothetical token totals and cache-hit ratios. They cover model usage only, in USD, and exclude Search, Fetch, taxes and third-party charges. Actual usage varies.
Light testing
Flash · off-peak · 1M input tokens · 0% cache hit · 0.5M output tokens
1 × $0.15 + 0.5 × $0.6
$0.45
Coding with repeated context
Flash · off-peak · 10M input tokens · 80% cache hit · 2M output tokens
8 × $0.003 + 2 × $0.15 + 2 × $0.6
$1.52
Larger agent workload
Pro · off-peak · 100M input tokens · 50% cache hit · 20M output tokens
50 × $0.022 + 50 × $0.66 + 20 × $1.98
$73.70
For comparison, the light-testing example costs $0.90 at Flash peak rates with the same token mix. These are examples, not fixed monthly fees.
Official API or a Coding Plan?
The official DeepSeek API may suit developers who want direct, usage-based billing. Our independent Coding Plans provide access through a managed gateway with a plan allowance and specific usage windows. Their credentials and terms are separate from the official API.
The best choice depends on your workload and the available plan. Token billing and plan allowances are not automatically equivalent; review current prices, limits, renewal terms and stock before choosing.
Frequently asked questions
How much does the DeepSeek API cost?
DeepSeek bills model usage by input and output tokens. The rate depends on the model, whether input is served from cache, and whether the request falls in a peak or off-peak period. See the table above for rates per million tokens.
Is the DeepSeek API free?
The official API is usage-based. Granted balance or promotional credits may be available under DeepSeek’s current terms, but free credits should not be assumed. Check your account and the official documentation before budgeting.
How much do one million tokens cost?
The answer depends on whether those tokens are input or output, the model, the time and, for input, cache hits. One million input tokens and one million output tokens are separate quantities with different rates.
What are the DeepSeek API peak hours?
Peak hours are 09:00–12:00 and 14:00–18:00 Beijing time (01:00–04:00 and 06:00–10:00 UTC), Monday–Friday, excluding Chinese public holidays. All other hours, weekends, and Chinese public holidays are off-peak.
Does cached input cost less?
Yes. DeepSeek lists a lower input rate for cache hits. Actual savings depend on how much input is served from cache; output is charged separately.
Is DeepSeek Flash cheaper than DeepSeek Pro?
In the rates shown above, Flash has lower input and output prices than Pro in both periods. Compare model capability and your actual token mix as well as the price.
How is the official API different from a Coding Plan?
The official DeepSeek API deducts usage charges from an API balance. Our independent Coding Plans provide access through a separate managed gateway with plan-specific prices, usage limits and billing terms. They are different services.