Search across 332 pages

Try a tool name, category, or "lifetime deal"

DeepSeek logo

DeepSeek

AI-first

AI Chat & Assistants

DeepSeek is a China-based AI lab shipping an unlimited free chat product and a pay-as-you-go API priced well below Western frontier-lab equivalents.

ABOUT DEEPSEEK

DeepSeek runs an unlimited free consumer chat product and prices its API several orders of magnitude below OpenAI or Anthropic's equivalents, particularly with cache hits factored in. The published model weights (V4 line) run self-hosted.

OVERVIEW

Who is it for?
Developers, technical users, and self-hosters needing low-cost reasoning and coding models via chat, API, or open weights.
Problem
Frontier chat and API pricing from Western labs is prohibitive for high-volume coding and reasoning workloads, and closed models cannot be self-hosted for privacy or air-gapped use.
Solution
Unlimited free consumer chat and a pay-as-you-go API priced well below Western frontier-lab equivalents, with a 1M token context, thinking and non-thinking modes, and OpenAI- and Anthropic-compatible endpoints.
What makes it unique
Open-weight V4 models that run self-hosted, API pricing several orders of magnitude below OpenAI or Anthropic with aggressive cache-hit discounts, and dual OpenAI- and Anthropic-compatible endpoints for drop-in integration.

KEY FEATURES

  • 1M token context window
  • 384K maximum output tokens
  • Thinking and non-thinking modes
  • OpenAI-compatible API endpoint
  • Anthropic-compatible API endpoint
  • JSON output mode
  • Tool calls
  • Responses API
  • Chat prefix completion (beta)
  • FIM completion for code

PRICING PLANS

Every published plan tier. Editorial reasoning about which one to buy lives on our reviews and best-of pages, not here.

Chat (chat.deepseek.com)

Free

Free web and mobile chat interface with thinking mode and search.

Seats: 1

API deepseek-v4-flash

Contact sales

Off-peak $0.22 per 1M input tokens (cache miss), $0.007 cache hit, $0.66 per 1M output tokens; peak rates double.

API deepseek-v4-pro

Contact sales

Off-peak $0.66 per 1M input tokens (cache miss), $0.022 cache hit, $1.98 per 1M output tokens; peak rates double.

COMPANY

Founded
2023
Headquarters
Hangzhou, Zhejiang, China
Founders
Liang Wenfeng
Employees
101-200
Funding
$7B Series A (May 2026)
Publicly traded
No

Where ZPlatform covers DeepSeek

Covered in

ZPlatform pages that treat this tool as their subject.

Sources

  1. DeepSeek API pricing - Pricing page · checked Jul 13, 2026 · may be stale for this field (48 days)

This profile answers what is DeepSeek?. For is it good? see any ZPlatform review. For which tool should I choose? see the relevant best-of ranking.