Skip to content
DIGINTOAI / MODEL

DeepSeek.

DeepSeek · Company articles ↗

DeepSeek's model family for chat, reasoning and coding agents. The API currently serves DeepSeek-V4.1-Flash-1017 as deepseek-flash and DeepSeek-V4-Pro-0813 as deepseek-v4-pro, and DeepSeek now also sells its own Search and Fetch APIs for live web search and page fetching.

Tools and API platforms

About this model

DeepSeek's language model family for conversation, reasoning and code tasks.

Official access and developer resources are listed below. Consult the provider for available models, regions and access requirements.

Plans and other pricing

DeepSeek charges by peak and off-peak hours; off-peak is always half price. Peak: deepseek-flash (V4.1-Flash-1017) $0.30/$1.20 per 1M, V4-Pro $1.32/$3.96; China ¥2/¥8 and ¥9/¥27. Both: 1M context, 384K max out. Search/Fetch $0.75 (¥5)/1K.

deepseek-v4-pro input

Usage based1.32 USD / per 1M input tokens
Region
Global
Conditions
V4-Pro-0813 peak price; cache hit $0.044; off-peak $0.66 (cache $0.022); peak=UTC Mon–Fri 01–04 & 06–10, excl. Chinese public holidays
Source
api-docs.deepseek.com ↗
Checked
· Last checked at least 30 days ago. Confirm the latest details with the provider.

deepseek-v4-pro output

Usage based3.96 USD / per 1M output tokens
Region
Global
Conditions
Peak price; off-peak $1.98
Source
api-docs.deepseek.com ↗
Checked
· Last checked at least 30 days ago. Confirm the latest details with the provider.

deepseek-flash input

Usage based0.30 USD / per 1M input tokens
Region
Global
Conditions
V4.1-Flash-1017 peak price; cache hit $0.006; off-peak $0.15 (cache $0.003); legacy deepseek-v4-flash / deepseek-v4-flash-vision-exp names bill at this price
Source
api-docs.deepseek.com ↗
Checked
· Last checked at least 30 days ago. Confirm the latest details with the provider.

deepseek-flash output

Usage based1.20 USD / per 1M output tokens
Region
Global
Conditions
Peak price; off-peak $0.60
Source
api-docs.deepseek.com ↗
Checked
· Last checked at least 30 days ago. Confirm the latest details with the provider.

deepseek-v4-pro input

Usage based9 CNY / per 1M input tokens
Region
Mainland China
Conditions
V4-Pro-0813 peak price; cache hit ¥0.30; off-peak ¥4.5 (cache ¥0.15); peak=Beijing time weekdays 9–12 & 14–18, excl. public holidays
Source
api-docs.deepseek.com ↗
Checked
· Last checked at least 30 days ago. Confirm the latest details with the provider.

deepseek-v4-pro output

Usage based27 CNY / per 1M output tokens
Region
Mainland China
Conditions
Peak price; off-peak ¥13.5
Source
api-docs.deepseek.com ↗
Checked
· Last checked at least 30 days ago. Confirm the latest details with the provider.

deepseek-flash input

Usage based2 CNY / per 1M input tokens
Region
Mainland China
Conditions
V4.1-Flash-1017 peak price; cache hit ¥0.04; off-peak ¥1 (cache ¥0.02)
Source
api-docs.deepseek.com ↗
Checked
· Last checked at least 30 days ago. Confirm the latest details with the provider.

deepseek-flash output

Usage based8 CNY / per 1M output tokens
Region
Mainland China
Conditions
Peak price; off-peak ¥4
Source
api-docs.deepseek.com ↗
Checked
· Last checked at least 30 days ago. Confirm the latest details with the provider.

Search API

Usage based0.75 USD / per 1,000 successful requests
Region
Global
Conditions
Billed only for HTTP 200 with results; server-side calls via DeepSeek Harness, Claude Code or Codex free until 2026-11-01 00:00 Beijing time
Source
api-docs.deepseek.com ↗
Checked
· Last checked at least 30 days ago. Confirm the latest details with the provider.

Fetch API

Usage based0.75 USD / per 1,000 successful requests
Region
Global
Conditions
Billed only for HTTP 200 with results; server-side calls via DeepSeek Harness, Claude Code or Codex free until 2026-11-01 00:00 Beijing time
Source
api-docs.deepseek.com ↗
Checked
· Last checked at least 30 days ago. Confirm the latest details with the provider.

Search API

Usage based5 CNY / per 1,000 successful requests
Region
Mainland China
Conditions
Billed only for HTTP 200 with results; server-side calls via DeepSeek Harness, Claude Code or Codex free until 2026-11-01 00:00 Beijing time
Source
api-docs.deepseek.com ↗
Checked
· Last checked at least 30 days ago. Confirm the latest details with the provider.

Fetch API

Usage based5 CNY / per 1,000 successful requests
Region
Mainland China
Conditions
Billed only for HTTP 200 with results; server-side calls via DeepSeek Harness, Claude Code or Codex free until 2026-11-01 00:00 Beijing time
Source
api-docs.deepseek.com ↗
Checked
· Last checked at least 30 days ago. Confirm the latest details with the provider.

Links and prices are maintained by our editors. Availability, limits and final charges are determined by the provider.

API cost calculator

API usage is separate from chat subscriptions and Coding Plans.

deepseek-flash

DeepSeek API · USD peak · Supported first-party API regions, default routing

Input / 1M tokens: 0.30 USD
Output / 1M tokens: 1.20 USD

Account access and model availability are determined by the provider. Subscription plans do not offset these API charges.

Standard synchronous text with uncached input; output includes billed thinking tokens. Excludes caching, batch, fast/priority, regional premiums, media, tools and taxes. USD peak-rate budget: Mon–Fri 01–04 and 06–10 UTC, excluding Chinese public holidays. Other hours are half price. All calls here use peak rates; there is no automatic time allocation. Conservative combined/output caps: 1,000,000/384,000 tokens.

Verified: 2026-11-18

Official pricing ↗

Estimate API costs →

deepseek-v4-pro

DeepSeek API · USD peak · Supported first-party API regions, default routing

Input / 1M tokens: 1.32 USD
Output / 1M tokens: 3.96 USD

Account access and model availability are determined by the provider. Subscription plans do not offset these API charges.

Standard synchronous text with uncached input; output includes billed thinking tokens. Excludes caching, batch, fast/priority, regional premiums, media, tools and taxes. USD peak-rate budget: Mon–Fri 01–04 and 06–10 UTC, excluding Chinese public holidays. Other hours are half price. All calls here use peak rates; there is no automatic time allocation. Conservative combined/output caps: 1,000,000/384,000 tokens.

Verified: 2026-11-18

Official pricing ↗

Estimate API costs →