iilib
API

DeepSeek API

DeepSeek API serves its current V4 models through OpenAI-compatible and Anthropic-compatible interfaces and separately offers Search and Fetch JSON endpoints for live web content.

API documentationOfficial siteVerified Nov 15, 2026
Provider type
First Party
Provider
DeepSeek
Connected models
2
Product integrations
7
OpenAI compatible
Yes

Overview

DeepSeek API supports chat completions, Responses, tool use, JSON, thinking controls and FIM. The deepseek-flash alias serves the October 17 V4.1-Flash-1017 update. Standalone Search (text only; image search is currently available via model web tools) and Fetch are billed separately per successful result-bearing request; these are not token inference prices.

API details

Base URL
https://api.deepseek.com
Authentication
Api Key
Styles
Rest, Sdk
Pricing model
Usage Based
Official SDKs
—

Capabilities

Chat completionsResponses APITool callsJSON outputThinking modesFIM completionAnthropic-compatible APISearch API: live text web result lookup; image search via model web toolsFetch API: web page content as Markdown

Model tokens vary by tier and peak/off-peak window. Search and Fetch each cost USD 0.75 per 1,000 successful result-bearing requests, including server-side calls through DeepSeek Harness, Claude Code and Codex. Failed, empty-result and timed-out calls are not billed. Image search is currently available through model web tools, not the standalone Search API. Tool fees are separate from model token charges.

Models currently mapped to this API

This is the verified iilib subset, not the provider's complete model catalog.

Tools and agents integrating this API

Only integrations backed by explicit product or provider documentation are shown.

Provider