Tool profile · Assistants

DeepSeek

A general model platform for web chat and API integration

DeepSeek provides web chat and APIs compatible with OpenAI and Anthropic formats. As verified on 2026-08-08, its main API models are DeepSeek-V4-Flash and DeepSeek-V4-Pro, both with 1M context and thinking or non-thinking modes.

AssistantReasoningCodingAPIAgents

Visit official site Back to directory

01 / Overview

About

DeepSeek provides both consumer web chat and developer APIs. The current V4 family includes V4-Flash for speed and cost efficiency and V4-Pro for complex reasoning and agent work; official docs describe both as 1M-context models with thinking and non-thinking modes.

02 / Core features

Core features

Web chat

Use official chat access for questions, writing, reasoning, and coding with V4 models.

Two V4 models

V4-Flash targets fast, economical calls while V4-Pro targets more complex reasoning and agent tasks.

Long context and dual modes

Both API models support 1M context and can switch between thinking and non-thinking modes.

Developer APIs

Offers OpenAI Chat Completions and Anthropic-compatible interfaces with JSON output and tool calling.

03 / Use cases

Use cases

Everyday assistance and content

Use web chat for organization, writing, translation, and structured output.

Complex reasoning and long documents

Use long context for large references, technical documents, and multi-step problems.

Coding and agent workflows

Power code generation, tool calling, and automated agents.

Product integration

Connect DeepSeek to applications built around OpenAI- or Anthropic-style APIs.

04 / Platforms and languages

Platforms and languages

Platforms and devices

WebOpenAI-compatible APIAnthropic-compatible API

Languages

ChineseEnglishand other natural languages

The API base URL is https://api.deepseek.com. Check official docs for current models, concurrency, and interface capabilities.

05 / Pricing and quotas

Pricing and quotas

API usage is billed by tokens. The prices below come from the official Chinese and English pricing pages as verified on 2026-08-08; recheck before funding or integration because models and prices can change.

DeepSeek-V4-Flash

$0.14 / 1M uncached input tokens / $0.0028 cache hit; $0.28 / 1M output tokens

1M context with thinking/non-thinking modes, JSON output, and tool calls; the official concurrency limit is currently 2,500.

DeepSeek-V4-Pro

$0.435 / 1M uncached input tokens / $0.003625 cache hit; $0.87 / 1M output tokens

1M context with thinking/non-thinking modes, JSON output, and tool calls; the official concurrency limit is currently 500.

06 / Similar products

Common questions

01How do V4-Flash and V4-Pro differ?

Flash emphasizes speed and price efficiency, while Pro targets complex reasoning and agent work; both support 1M context and thinking or non-thinking modes.

02Can existing OpenAI SDK integrations be reused?

DeepSeek provides an OpenAI Chat Completions-compatible base URL and also an Anthropic-format API.

03Do web chat and API use the same billing?

API usage is billed by tokens; check the current chat product for its own availability and limits.

08 / Sources

Sources

SourceDeepSeek official website

Web product entry and company identity, verified 2026-08-08.

2026-08-08 SourceDeepSeek V4 official release

V4-Flash, V4-Pro, web access, 1M context, and API compatibility.

2026-08-08 SourceDeepSeek official models and pricing

Current models, capabilities, concurrency, and pricing, verified 2026-08-08.

2026-08-08 SourceDeepSeek Chat Completions API

Request format, model parameters, and tool-calling support.

2026-08-08 SourceWatcha product page

Directory context only; not used as primary evidence.

2026-07-31