Back to Tools
DeepSeek

DeepSeek

Deepseek

AI lab whose models are available in a chat app and through a pay-per-token API compatible with OpenAI and Anthropic request formats.

Updated 19d ago
FreemiumChatbots⚡ 55

About DeepSeek

DeepSeek offers large language models through an API and through its own app and website, and its developer documentation is the most detailed public description of what is on offer. The quick-start page explains that the API uses a format compatible with the OpenAI and Anthropic APIs, so existing SDKs and tools can be pointed at DeepSeek by changing configuration. It also says popular coding assistants such as Claude Code, GitHub Copilot and OpenCode can use DeepSeek as their backend model directly, and it mentions a developer preview of an agent harness.

The models and pricing page lists two API models: deepseek-flash, served by DeepSeek-V4.1-Flash, and deepseek-v4-pro. Both support a thinking mode (the default) and a non-thinking mode, JSON output, tool calls, a Responses API and an Anthropic-format API, with a maximum output of 384K tokens. The page marks vision as supported on Flash but not on Pro, and prefix and fill-in-the-middle completion as beta features. A release note adds that reasoning effort can be set to low, high or max, and that V4 Pro is reachable in the app and on the web through "Expert Mode".

Billing is per million tokens and depends on the time of day. Peak hours are 01:00–04:00 and 06:00–10:00 UTC on weekdays, excluding Chinese public holidays; everything else counts as off-peak, at half the peak rate. Rates differ between the two models and between input on a cache miss, input on a cache hit (much cheaper) and output, and the page lists concurrency limits of 2,500 for Flash and 500 for Pro.

The page says prices may vary and that DeepSeek reserves the right to adjust them, and model names have already been retired and remapped once. Before integrating, confirm the current model names, whether you need vision (listed for Flash only) and how peak-hour pricing affects your workload.

Sources

The official pages this summary was checked against. Details such as pricing change; confirm on the vendor site.

Key Information

Pricing
Freemium
Platforms
Web, iOS, Android, API
Category
Chatbots
Developer
Deepseek
Last Updated
19d ago
Verified
Yes
Website Status
Online
Documentation
View Docs

Features

API compatible with OpenAI and Anthropic request formats
Two API models listed: deepseek-flash and deepseek-v4-pro
Thinking and non-thinking modes; JSON output and tool calls supported
Per-million-token pricing with lower off-peak rates
Vision supported on the Flash model; not on Pro (per the docs)
Also available in a chat app and on the web

Mentioned in Our Blog