Menu

Guides · dev · deepseek api key

DeepSeek API key: top up, then use the off-peak hours on purpose

Published 16 Oct 2026. Updated 16 Oct 2026. Researched 16 Oct 2026.

A DeepSeek API key is a balance, not a free tier and not a coding-plan subscription. The platform is platform.deepseek.com. We did not create a key. The steps below are the shape of the docs, not a screenshot tour.

From zero to a priced call1Open the DeepSeek API platform2Create a key and copy it once3Top up only what you will spend this month4Call deepseek-flash at the OpenAI or Anthropic base URL5Read the bill and see whether you landed in peak or off-peak

Base URLs printed on the pricing page:

https://api.deepseek.com
https://api.deepseek.com/anthropic

Model names: deepseek-flash for V4.1 Flash, deepseek-v4-pro for the pro column. Peak cache-miss prices are $0.30 / $1.20 for flash and $1.32 / $3.96 for pro, per million tokens. Off-peak halves both. Cache hits are a separate, much smaller input price on the same page.

Illustrative Coding-agent session costDeepSeek-V4.1-Flash$0.178DeepSeek-V4-Pro-0813$0.752Anthropic: Claude Opus 5.5$2.94OpenAI: GPT-6.1 Sol$1.47
Coding-agent session: 3 million input tokens, 90% served from cache, 60 thousand output tokens. Cache-write fees are not included. Chinese rows use the lab published rate (DeepSeek uses the peak rate). Comparators are OpenRouter list prices for Anthropic: Claude Opus 5.5 ($8.00/M blended) and OpenAI: GPT-6.1 Sol. Snapshot 2026-10-16. Cache-write fees are excluded. A missing cache price is billed as a full input token.
ModelSessionvs Opusvs SolCache price
DeepSeek-V4.1-Flash$0.17894% cheaper88% cheaperpublished
DeepSeek-V4-Pro-0813$0.75274% cheaper49% cheaperpublished
Anthropic: Claude Opus 5.5$2.94——published
OpenAI: GPT-6.1 Sol$1.47——published

The percentages in this table compare session cost. The price table's percentages compare blended token price. They answer different questions.

Payment from abroad

The docs say you top up. They do not, in the section we used, give us a card-brand matrix we can reprint. The payment guide keeps that as an open question instead of inventing Visa or PayPal. If the platform rejects your card, OpenRouter is the workaround, at the peak-equivalent flash price rather than off-peak. That trade is in OpenRouter versus direct.

Do not put the key in a screenshot, a Discord, or a client-side app. The usual leakage story is unchanged because the vendor is Chinese.

Related: is the app free?, V4.1 Flash on OpenRouter.

Questions

Is the DeepSeek API free?

No. The pricing page bills a balance. The chat app is the free product.

What is off-peak?

Half of peak. Peak is 01:00-04:00 and 06:00-10:00 UTC, Monday to Friday, excluding Chinese public holidays.

Sources

  1. DeepSeek API pricing, accessed 2026-10-16
  2. DeepSeek platform, accessed 2026-10-16
  3. OpenRouter models API, accessed 2026-10-16