Skip to content
DeepSeek

DeepSeek

AI chat assistant and model API from DeepSeek, with open models.

Ready to get started?

Launch DeepSeek and start using it right away

About DeepSeek

DeepSeek is an AI company that offers a chat assistant at chat.deepseek.com and in a mobile app, and an API for developers. Chat is free to use. The API is billed per million tokens, with lower prices for cached input and for off-peak hours.

The current models are DeepSeek-V4.1-Flash and DeepSeek-V4-Pro. DeepSeek describes V4.1-Flash as improved for text and agent tasks and as having native multimodal vision. The API supports a thinking mode, tool use and a context window of one million tokens, and it is compatible with the OpenAI and Anthropic API formats, so it can be used from agent and coding tools.

DeepSeek publishes its models on Hugging Face.

Our take

Good for
low-cost AI chat and API access, and open models you can run yourself.
Skip it if
you would send sensitive business data to its hosted app or API.
The catch
the hosted service keeps data on servers in China and regulators have questioned it, so self-hosting the open models is the way to keep data under your control.

Checked . Plans and limits change, so confirm them on the vendor's site before you buy.

Key Features

A free chat assistant on the web and in a mobile app.
A thinking mode for harder reasoning tasks.
Tool use and agent support through the API.
A context window of one million tokens on both API models.
An API compatible with the OpenAI and Anthropic formats.
Lower API prices for cached input and for off-peak hours.
Models published on Hugging Face.

Use Cases

Ask questions and draft text in the free chat assistant.
Work through a coding or maths problem using thinking mode.
Add a low-cost model to an app through the API.
Run agent or coding tools that connect through the OpenAI or Anthropic API formats.
Analyse a long document within the one million token context window.