AI pricing · November 2026
DeepSeek API pricing
| Provider | Cached in | Context | ||||
|---|---|---|---|---|---|---|
| DeepSeek V4.1 Flash peak-hour rates (01:00–04:00 and 06:00–10:00 UTC on weekdays, cache miss); off-peak half price: $0.15 in / $0.60 out / $0.003 cached | DeepSeek | $0.300 | $1.20 | $0.006 | 1M | $5.40 |
| DeepSeek V4 Pro peak-hour rates (01:00–04:00 and 06:00–10:00 UTC on weekdays, cache miss); off-peak half price: $0.66 in / $1.98 out / $0.022 cached | DeepSeek | $1.32 | $3.96 | $0.044 | 1M | $21.12 |
Cached input and batch discounts
| Model | Input $/1M | Cached input $/1M | Cache saving | Batch |
|---|---|---|---|---|
| DeepSeek V4.1 Flash | $0.30 | $0.006 | 98% off | — |
| DeepSeek V4 Pro | $1.32 | $0.044 | 97% off | — |
What a typical workload costs
An app that sends 1,000 requests a day, each with a 2,000-token prompt and a 500-token answer, uses 60M input and 15M output tokens in 30 days. If 1,500 of those prompt tokens are the same system prompt and context every time, prompt caching bills 45M of them at the cached rate (cache writes and storage not included).
| Model | Per month | With caching |
|---|---|---|
| DeepSeek V4.1 Flash | $36.00 | $22.77 |
| DeepSeek V4 Pro | $139 | $81.18 |
Closest-priced models from other providers
Matched on a 3:1 mix of input to output tokens. Open-model hosts are left out.
| DeepSeek model | In / out $/1M | Closest elsewhere |
|---|---|---|
| DeepSeek V4.1 Flash | $0.30 / $1.20 | Gemini 3.1 Flash-Lite (Google) $0.25 / $1.50 · Codestral (Mistral) $0.30 / $0.90 |
| DeepSeek V4 Pro | $1.32 / $3.96 | Claude Haiku 4.5 (Anthropic) $1 / $5 · GPT-6.1 Terra (OpenAI) $1 / $5 |
Source: DeepSeek pricing, checked 2026-11-02. To price a real prompt, paste it into the token counter.
FAQ
How much does the DeepSeek API cost?+
DeepSeek charges per token. On 2026-11-02 its official pricing page listed, per 1M tokens: DeepSeek V4.1 Flash $0.30 input / $1.20 output and DeepSeek V4 Pro $1.32 input / $3.96 output.
What is the cheapest DeepSeek model?+
DeepSeek V4.1 Flash: $0.30 per 1M input tokens and $1.20 per 1M output tokens. For the example workload on this page (30,000 requests a month) that is about $36.00 a month, against $139 on DeepSeek V4 Pro, the most expensive DeepSeek model here.
Does DeepSeek have prompt caching or cheaper cached input?+
Yes. Cached input tokens cost $0.006 on DeepSeek V4.1 Flash (98% less than regular input) and $0.044 on DeepSeek V4 Pro (97% less than regular input).
Does DeepSeek offer a batch discount?+
This page has no batch price for DeepSeek models; check the official pricing page linked below for asynchronous or off-peak options.
Is the DeepSeek API cheaper off-peak?+
Yes, half price. The prices in the table are peak rates, which apply 01:00–04:00 and 06:00–10:00 UTC, Monday to Friday (except Chinese public holidays). All other hours, weekends and those holidays are billed at half the peak rate, for cache hits, cache misses and output alike.
How does DeepSeek V4.1 Flash compare on price with other AI APIs?+
At $0.30 input / $1.20 output per 1M tokens, the closest-priced models from other providers are Gemini 3.1 Flash-Lite (Google, $0.25 / $1.50) and Codestral (Mistral, $0.30 / $0.90). Price per token isn't the whole bill: compare quality on your own prompts, and how many tokens each model uses for the same task.
How current are these prices?+
They were copied from DeepSeek's official pricing page on 2026-11-02, and the month in this page's title is that check date. Prices change often, so confirm on the official page before you commit to a budget.
Other providers
Claude API pricing · OpenAI API pricing · Gemini API pricing · Grok API pricing · Mistral API pricing · Groq API pricing · Together AI API pricing · Fireworks AI API pricing · All AI model prices compared