p2d
> open deepseek --ru_

DeepSeek in Russia without a VPNDeepSeek V4.1 Flash, V4 Pro, V3.2, and R1 in the P2D AI aggregator — no subscription

A Chinese model with strong coding and reasoning — and one of the most affordable: a typical response costs a fraction of a cent. Write in Russian, top up your balance via SBP.

  • DeepSeek V4.1 Flash$0.00022
  • DeepSeek V4 Pro$0.0041
  • DeepSeek V4 Flash$0.0008
  • DeepSeek V3.2$0.0011
p2d://ai-agregator/deepseek

> you: Write a Python function that removes duplicates from a list and preserves order

deepseek-v4-pro ▸ def unique(items): seen = set() return [x for x in items if not (x in seen or seen.add(x))]

[ok] answer ready · charged $0.0041 · $10 gets you ≈ 2,443 such answers

The terminal shows four tasks for DeepSeek models: math, code, editing, and a logic puzzle. After each response you can see what it cost.
> ls ./deepseek

DeepSeek models in the aggregator

The aggregator has 14 DeepSeek models — from V3 to V4.1 Flash. The main ones:

  1. September 2026

    DeepSeek V4.1 Flashnew

    A fast model on DeepSeek's new architecture. Understands text and images, has a context of up to 1M tokens, and costs a fraction of a cent per response.

    context
    1M
    regular response
    $0.00022
    for $10
    ≈ 44,523
  2. August 2026

    DeepSeek V4 Pro

    The large DeepSeek V4 Pro model (final release). For complex code, reasoning, and long documents.

    context
    1M
    regular response
    $0.0041
    for $10
    ≈ 2,443
  3. July 2026

    DeepSeek V4 Flash

    An updated version of V4 Flash: an economical model with up to 1M tokens of context.

    context
    1M
    regular response
    $0.0008
    for $10
    ≈ 12,426
  4. December 2025

    DeepSeek V3.2

    A 2025 model focused on efficiency, reasoning, and tool use.

    context
    163K
    regular response
    $0.0011
    for $10
    ≈ 8,710
  5. May 2025

    DeepSeek R1

    DeepSeek R1 reasoning model (update of May 28, 2025): thinks before answering, strong in logic and math.

    context
    163K
    regular response
    $0.004
    for $10
    ≈ 2,493
> deepseek --help

What DeepSeek can do

Code

Writes and explains code in Python, JavaScript, and other languages, finds bugs, and suggests improvements.

> Find the bug in this script

Math and logic

Solves problems with an explanation of the steps, checks calculations.

> Solve the system of equations and check the answer

Texts

Writes, shortens and edits texts, translates into other languages.

> Turn this text into three posts

High-volume tasks

Its low price makes it suitable for processing thousands of similar requests.

> Label 1000 reviews: positive or negative
> price deepseek --beta 1.0

How much does DeepSeek cost

No subscription: pay only for responses. The price is from $0.00022 to $0.0041 per typical response.

Beta 1.0 prices may change

Response size

≈ a 1000-token request and a 700-token answer — a typical conversation

Prices
ModelRequest,
$ per 1M tokens
Response,
$ per 1M tokens
ContextOne response$10 is enough for
DeepSeek V4.1 Flash 1017$0.3$1.21M$0.0011≈ 8,771
DeepSeek V4.1 Flash$0.0566$0.241M$0.00022≈ 44,523
DeepSeek V4 Flash Vision Exp$0.4312$1.29361M$0.0013≈ 7,480
DeepSeek V4 Pro 0813$1.32$3.961M$0.0041≈ 2,443
DeepSeek V4 Flash 0731$0.2596$0.77881M$0.0008≈ 12,426
DeepSeek V4 Pro 0423$0.7503$1.50061M$0.0018≈ 5,553
DeepSeek V4 Flash 0423$0.182$0.3641M$0.00044≈ 22,893
DeepSeek V3.2$0.56$0.84163K$0.0011≈ 8,710
DeepSeek V3.2 Exp$0.54$0.82163K$0.0011≈ 8,976
DeepSeek V3.1 Terminus$0.54$2163K$0.0019≈ 5,154
DeepSeek V3.1$0.5$1.9163K$0.0018≈ 5,464
R1 0528$1$4.3163K$0.004≈ 2,493
DeepSeek V3 0324$0.58$2.28163K$0.0022≈ 4,595
DeepSeek V3$0.5148$2.0574163K$0.002≈ 5,115

The estimate is approximate: the cost of a response depends on the length of your request and the model's answer. The actual generation cost is deducted from your balance.

> cat ./how-to-start.txt

How to use DeepSeek in Russia

  1. 01

    Sign up for P2D

    One account opens all the aggregator's models — no need to register with each one separately.

  2. 02

    Top up your balance via SBP

    Balance in dollars. No foreign card or VPN needed, no monthly fee.

  3. 03

    Choose DeepSeek

    Open the AI aggregator, choose a model, and ask your question in Russian. The cost of the result is deducted from your balance.

Sign up and get started →
> man deepseek

FAQ about DeepSeek

How to use DeepSeek in Russia?

Sign up for P2D, top up your balance via SBP, and choose DeepSeek in the AI aggregator. No VPN or foreign card needed.

Do you need a VPN for DeepSeek?

No. The P2D aggregator opens from Russia without a VPN, and the service itself sends the requests to the model.

How much does DeepSeek cost?

No subscription — you pay for each response. A typical DeepSeek V4.1 Flash response costs about $0.00022, so $10 covers about 44,523 such responses. Prices for all models are in the table on this page.

Is there a subscription for DeepSeek?

No subscription: you top up your balance in dollars and spend it only on what you generate. The same balance works for all of the aggregator's models.

How does DeepSeek V4 Pro differ from V4.1 Flash?

Pro is a large model for complex tasks, Flash is fast and cheap. Flash is usually enough for everyday questions and text.

Is there DeepSeek R1?

Yes, the aggregator has the reasoning model R1 (version 0528) — it thinks before answering and handles logic and math well.

Why is DeepSeek so cheap?

DeepSeek uses an architecture in which only part of the model runs for each response, so generation is cheaper. Prices for all versions are in the table above.

Does DeepSeek understand Russian?

Yes. Write in Russian — DeepSeek will answer in Russian.

Other models in the aggregator

> start --deepseek

Try DeepSeek right now

Sign up in a minute, top up via SBP, pay only for the result.

Start with DeepSeek →