AI Model Compare / DeepSeek

DeepSeek: pricing, context window and best use

DeepSeek by DeepSeek. Current model: DeepSeek V4.1 / V4 Pro. Best for: Value / Reasoning / Chinese.

MakerDeepSeek
Current modelDeepSeek V4.1 / V4 Pro
API input$0.15–0.66 / 1M tokens
API output$0.6–1.98 / 1M tokens
Context window1M
Free tierFree web app
Best forValue / Reasoning / Chinese
Official sitechat.deepseek.com ↗

What DeepSeek costs

API pricing is $0.15–0.66 per million input tokens and $0.6–1.98 per million output tokens. Subscription options: Web app: completely free · API Flash: $0.3 in / $1.2 out · API Pro: $1.32 in / $3.96 out. These are list prices — caching, batch discounts and reseller contracts change the real bill. Figures were checked against official pages in September 2026.

Context window and limits

DeepSeek has a 1M context window. That ceiling covers everything in one request: system prompt, documents, retrieved passages and the conversation so far. Beyond it the input is truncated, so long jobs are usually split into chunks with a running summary.

What DeepSeek is best at

DeepSeek is built for Value / Reasoning / Chinese. Because billing is per token, the effective cost depends on the input-to-output ratio: reading a lot and answering briefly is far cheaper than generating long output.

How DeepSeek compares

DeepSeek sits at the middle of the price range covered here. The nearest alternatives by input price are Llama ($0.08–0.27/1M in) and ChatGPT ($0.2–4/1M in).

FAQ

How much does DeepSeek cost per million tokens?

$0.15–0.66 for input and $0.6–1.98 for output, based on the list prices tracked on this page.

What is the context window of DeepSeek?

1M in a single request.

Is DeepSeek free to use?

Free web app. DeepSeek publishes current terms at chat.deepseek.com.

Other models on this page

Prices, context windows and plan limits are collected from each vendor's official pages and re-checked monthly. Last update: September 2026. This page is independent and not affiliated with DeepSeek.