DeepSeek · open weights

DeepSeek V4 Pro Price

How much does DeepSeek V4 Pro cost? On DeepSeek, DeepSeek V4 Pro costs $1.32 per 1 million input tokens and $3.96 per 1 million output tokens. A support chatbot using 20M input and 5M output tokens a month would pay about $46.20 a month at list price.

Read on the official api-docs.deepseek.com pricing page ↗ · checked 2026-11-21.
Input
$1.32
per 1M tokens
Output
$3.96
per 1M tokens
Blended
$1.98
3 input : 1 output
Cached input
$0.044
per 1M tokens, as listed
Context window
1,000K
tokens, as listed

Price Tiers

As listed on the official pricing page · USD per 1M tokens · checked 2026-11-21

TierWhen it appliesInputOutputCached input
standardPeak hours: 01:00–04:00 and 06:00–10:00 UTC, Monday–Friday$1.32$3.96$0.044
off peakAll other hours, weekends and Chinese public holidays$0.66$1.98$0.022

The whole request is billed at a tier’s rate once its prompt passes the limit. Batch prices apply to asynchronous batch jobs.

Price History

Every price we saw on the official page, with the date

Seen onInputOutputSource
2026-11-21$1.32$3.96pricing page ↗

What a Month Costs

List prices, no discounts

Support chatbot · 20M in, 5M out$46.20
Summarize documents · 60M in, 6M out$102.96
1M in and 1M out$5.28

Open the full calculator

How DeepSeek V4 Pro Compares

Computed from the 42 models on our price board, list prices checked 2026-11-21

Against the Whole Board

Of the 42 models on our price board, DeepSeek V4 Pro has the 15th lowest output price and the 19th lowest input price. Its output price of $3.96 per 1M tokens is 55% below the median of $8.75, and its input price of $1.32 is 25% below the median of $1.75. On the blended measure (3 input tokens for every output token, typical for chat), it costs $1.98 per 1M tokens.

Within DeepSeek’s Line-Up

We track 2 models from DeepSeek. By blended price, DeepSeek V4 Pro is the 2nd cheapest. The cheapest is DeepSeek V4.1 Flash at $0.30 in and $1.20 out; the most expensive is DeepSeek V4 Pro at $1.32 in and $3.96 out per 1M tokens. Moving down to DeepSeek V4.1 Flash would cut a blended bill by about 73%, if it handles your task well enough.

Input Versus Output

DeepSeek V4 Pro charges 3.0 times more for output tokens than for input tokens; across the board the median gap is 5.0 times. Apps that write long answers (drafting, code generation) spend most of their budget on output, so the output price matters more for them; apps that read a lot and answer briefly (classification, search, summaries) are driven by the input price.

Cost of Everyday Tasks

At list price, 1,000 chatbot replies (about 500 input and 300 output tokens each) cost $1.85. Summarising a 50-page report (about 25,000 input tokens) into a one-page answer (about 1,000 output tokens) costs $0.037 per document, so 1,000 such reports cost $36.96.

Prompt Caching

DeepSeek lists a cached-input price of $0.044 per 1M tokens for DeepSeek V4 Pro, 97% less than normal input. If an app resends the same 2,000-token instructions with 100,000 requests a month, those instructions cost $264 uncached and about $8.80 when they are served from the cache (cache-write fees, where a provider charges them, are not included).

Context Window

DeepSeek V4 Pro accepts up to 1,000,000 tokens of context, against a median of 1,000,000 for the models where a context size is listed. Long prompts are billed in full every time they are sent, and some providers charge a higher rate above a certain prompt length; check the official pricing page if your prompts are very long.

Price Changes

We have not seen DeepSeek V4 Pro’s price change since we started reading DeepSeek’s pricing page. If it changes, the old price stays in the history above and the change appears on the Wire.

Open or Closed

DeepSeek V4 Pro has open weights, so other companies can host it and you can run it yourself. Prices from other hosts can differ a lot from DeepSeek’s; where we track another host, it appears in the side panel.

Type a name, a topic or a format. Enter opens the full results.