DeepSeek API · OpenAI-compatible

DeepSeek V4.1 Flash at $0.21 per 1M input tokens.

−30% against DeepSeek’s official price from your first request. One OpenAI-compatible key for DeepSeek and every other major model. Pay by card in USD and point your OpenAI SDK at a new base URL.

  • $1 free when you sign up
  • +10% on your first top-up, up to $100
  • Failed requests are free
1M token contextTool calling · JSON outputPrepaid, no subscription
up to −40%

DeepSeek V4.1 Flash · per 1M input tokens

Official $0.30

$0.18Tier 3

Tier 1 for new accounts: $0.21. The more you top up, the lower it goes.

  • DeepSeek V4 Pro$1.32$0.792−40%

What a top-up buys

$20 is about 23,800 chat replies on V4.1 Flash.

Pick a model and an amount. Each number is how many requests of that size your balance covers at Tier 1.

What you pay for

A lower bill, long outputs, one key.

−30%

Below DeepSeek’s price

V4.1 Flash bills at $0.21 input and $0.84 output per 1M tokens from the first request, against DeepSeek’s official $0.30 and $1.20. Cached input bills at the official $0.006.

393K

Long answers in one reply

V4.1 Flash reads up to 1M tokens and writes up to 393K in a single response, with reasoning, tool calling and JSON output.

1 key

Every major model family

The same key reaches Qwen, GLM, Kimi, Claude, GPT and Gemini. Switch models by changing one string, from one prepaid balance in USD.

How pricing tiers work

Every account starts at Tier 1. Top up more to move up.

Your account starts on Tier 1, with the discount below from the first request. At $1,000 in cumulative top-ups it moves to Tier 2, at $10,000 to Tier 3, on its own, and every later request bills at the lower rate.

Tier 1 · you start hereBelow $1,000 in top-ups

V4.1 Flash −30%, V4 Pro −20% against DeepSeek’s official price. Purchased credits last 12 months.

Tier 2$1,000+ in top-ups

V4 Pro drops to −30%. V4.1 Flash stays −35%.

Tier 3$10,000+ in top-ups

The lowest rates: −40% on V4.1 Flash, −40% on V4 Pro.

  • 20% or more
  • 11–19%
  • 10% or less

Swipe the table for Tier 2 and 3 →

Per 1M tokensDeepSeek officialTier 1Tier 2Tier 3
V4.1 Flash · input$0.30$0.21−30%$0.195−35%$0.18−40%
V4.1 Flash · output$1.20$0.84−30%$0.78−35%$0.72−40%
V4 Pro · input$1.32$1.056−20%$0.924−30%$0.792−40%
V4 Pro · output$3.96$3.168−20%$2.772−30%$2.376−40%

Discounts apply to input and output. Cached input bills at DeepSeek’s official price at every tier (V4.1 Flash $0.006, V4 Pro $0.044 per 1M). Official price: the model maker's list price. Tier 1–3: OceanToken's standing discounts by cumulative top-up, not limited-time offers. Large volume? Talk to sales.

Which DeepSeek

Three DeepSeek models on the same key.

Pick by model ID. All three take a 1M-token context and support reasoning, tool calling and JSON output.

Newest

V4.1 Flash (1017)

The October 17, 2026 release. Text and image input, up to 393K output tokens.

deepseek/deepseek-v4.1-flash-1017

Earlier release

V4.1 Flash (September)

The September 2026 release, for teams that want to keep its behaviour, up to 944K output tokens, same price as 1017.

deepseek/deepseek-v4.1-flash

Larger model

V4 Pro

DeepSeek’s larger model, text only, up to 393K output tokens. Use it when Flash isn’t enough for the task.

deepseek/deepseek-v4-pro

Setup

From sign-up to first request in a few minutes.

Step 1 · Sign in
Sign in to OceanToken Continue with Googleyou@company.com Continue with email
Step 2 · Add credits
Add credits
$5$20$50$100
Balance
$20.00
Tier 1 · −30% on Flash
≈ 23,800 V4.1 Flash chat replies
Step 3 · Change the base URL
# Your API key from the OceanToken console client = OpenAI( base_url="https://api.oceantoken.ai/v1", api_key="YOUR_API_KEY") client.chat.completions.create( model="deepseek/deepseek-v4.1-flash-1017", messages=[{"role": "user", "content": "Hi"}])
Only the base URL, key and model ID change. Full quickstart
Step 4 · Top up more, pay less
Tier 1Below $1,000Flash −30%
Tier 2$1,000+Pro −30%
Tier 3$10,000+−40%
Large volume? Talk to sales.

Questions

Before you top up

What does $20 get me?

On V4.1 Flash at Tier 1, about 23,800 chat replies of 2,000 input and 500 output tokens, about 15,400 agent steps, or about 880 long documents of 100,000 input and 2,000 output tokens. On V4 Pro, about 5,400 of the same chat replies.

How do your prices compare with DeepSeek’s own API?

Against DeepSeek’s official list price, V4.1 Flash is −30% at Tier 1 ($0.21 input and $0.84 output per 1M, against $0.30 and $1.20) and V4 Pro is −20% ($1.056 and $3.168, against $1.32 and $3.96). Cached input bills at DeepSeek’s official price.

Which DeepSeek models can I use?

V4.1 Flash (October 17, 2026 release), the September V4.1 Flash release, and V4 Pro. All have a 1M-token context, tool calling and structured output.

Is there a subscription?

No. You prepay a balance in USD by card, and each request is deducted from it. Purchased credits last 12 months.

When do I move to Tier 2?

When your cumulative top-ups reach $1,000; Tier 3 starts at $10,000. Gifts and credit grants count toward the total. Your account moves up on its own and every later request bills at the lower rate; the console shows the rate that applies to you. Expect large volume? Talk to sales.

What happens when my balance reaches $0?

New requests are refused until you add credits.

Do I need to change my code?

Only the base URL and the key. The API is OpenAI-compatible, and the Anthropic Messages format works too.

More on OceanToken

Every DeepSeek model, side by side.

Ready

Get a key and send your first DeepSeek request today.

Get API Key Prepaid · credits last 12 months
$0.21 / 1M inV4.1 Flash · −30%Get API Key