DeepSeek is a free chatbot from a Chinese AI lab that open sources its models and sells one of the cheapest AI APIs around. It is worth trying for general reasoning and coding questions, and its API suits cost conscious developers, but its privacy policy stores your data in China, so keep anything sensitive out of it.
What DeepSeek is
DeepSeek is an AI company focused on general purpose models, and it open sources many of them, including DeepSeek-V4 and the R1 reasoning model DeepSeek ↗. Its consumer product is a chatbot you can use on the web, in a mobile app or in a desktop app called DeepSeek Harness DeepSeek ↗.
The chat offers two headline modes: deep thinking, which reasons step by step before answering, and smart search, which looks things up on the web DeepSeek ↗. Its newest model at the time of checking is DeepSeek-V4.1-Flash, which DeepSeek says improves text and agent performance and adds native image understanding DeepSeek ↗.
Developers can also pay for the same models through the DeepSeek API DeepSeek API pricing ↗. Hands on testing is in progress.
What it does well
It is free to chat. Zapier picks DeepSeek as the best free and open source chatbot for reasoning, praising its highly intelligent R1 reasoning model Zapier ↗.
It is open. Because DeepSeek publishes its model weights, you can run them on your own hardware or a cloud server you control, which Zapier points out as the workaround for privacy concerns DeepSeek ↗ Zapier ↗. Few big chatbot makers allow that.
The API is very cheap. DeepSeek-V4.1-Flash costs $0.30 per million input tokens and $1.20 per million output tokens at peak times, and half that off peak DeepSeek API pricing ↗. Cached input costs as little as $0.003 per million tokens off peak DeepSeek API pricing ↗.
It handles long inputs. Both API models accept up to 1 million tokens of context and can return up to 384,000 tokens of output DeepSeek API pricing ↗. Both support thinking and non thinking modes DeepSeek API pricing ↗.
There is a training opt out. DeepSeek's policy gives users the right to opt out of having their data used to train its models DeepSeek privacy policy ↗.
Where it falls short
Your data goes to China. DeepSeek's privacy policy says it collects, processes and stores personal data in the People's Republic of China DeepSeek privacy policy ↗. That includes your text and voice input, uploaded files, photos and chat history, plus device data such as IP address and device identifiers DeepSeek privacy policy ↗.
Your chats can train its models by default. The policy says personal data is used to train its models, and the opt out is described as a right you can exercise, with privacy@deepseek.com given as the contact DeepSeek privacy policy ↗. Data may also be shared with law enforcement and public authorities when required by law DeepSeek privacy policy ↗.
Governments are wary of it. At least 17 US states ban DeepSeek on government devices and networks, though no law stops private use BGR ↗. Italy's data protection watchdog barred the chatbot, and South Korea removed it from app stores for a time BGR ↗.
It avoids some topics. Zapier notes the app does not discuss sensitive political topics such as Tiananmen Square, and that it occasionally hallucinates Zapier ↗.
Details are thin in English. The home page does not state usage limits for the free chat, and the API page does not mention a free API tier DeepSeek ↗ DeepSeek API pricing ↗.
Pricing and plans
Chat: free. Zapier lists the DeepSeek chatbot as free, and DeepSeek's site shows no paid chat plan DeepSeek ↗ Zapier ↗.
API, DeepSeek-V4.1-Flash: $0.15 per million input tokens on a cache miss and $0.60 per million output tokens off peak, doubling to $0.30 and $1.20 at peak DeepSeek API pricing ↗.
API, DeepSeek-V4-Pro: $0.66 per million input tokens on a cache miss and $1.98 per million output tokens off peak, doubling to $1.32 and $3.96 at peak DeepSeek API pricing ↗.
Peak hours are 01:00 to 04:00 and 06:00 to 10:00 UTC on weekdays, excluding Chinese public holidays. All other times are billed at half price DeepSeek API pricing ↗.
Self hosting: the open source models are free to download, but you pay for your own servers DeepSeek ↗ Zapier ↗.
Who should (and shouldn't) use it
- Good fit: anyone who wants a free reasoning chatbot for non sensitive questions, study or coding help.
- Good fit: developers who need a low cost API with a large context window.
- Good fit: teams that want to self host an open model on their own infrastructure.
- Skip it if: you would paste client records, financial details or confidential business information.
- Skip it if: you work for or with a government agency that bans it.
- Skip it if: you need neutral answers on Chinese political topics.
Alternatives worth a look
- ChatGPT: our pick for most people, with unlimited text chats, web search and projects on its free plan ChatGPT pricing ↗.
- Claude: a strong free choice for long writing and working with documents Claude pricing ↗.
- Duck.ai: DuckDuckGo's chatbot puts a privacy layer in front of well known AI models, and Zapier picks it for data privacy Zapier ↗ Duck.ai privacy ↗.
Where it appears
ComparedThe Best Free AI ChatbotWhat’s good
- Strong reasoning reputation
- Open models
- Developer platform
What to know
- Data stored in China per privacy policy
- Not suited to sensitive data
Key features
- Reasoning focused models
- Open source model releases
- Paid developer platform
Questions people ask
Is DeepSeek free?
Yes. The DeepSeek chatbot is free to use, and its open source models are free to download. The API is paid per token DeepSeek ↗ DeepSeek API pricing ↗ Zapier ↗.
Is DeepSeek safe to use?
Treat it as unsafe for sensitive data. Its privacy policy says personal data, including chat history and uploaded files, is processed and stored in China and may be used to train its models DeepSeek privacy policy ↗.
Is DeepSeek banned in the US?
Not for private use. At least 17 states ban it on government devices and networks, but no law stops residents or companies from using it BGR ↗.
How much does the DeepSeek API cost?
DeepSeek-V4.1-Flash costs $0.30 per million input tokens and $1.20 per million output tokens at peak, and half that off peak. V4-Pro costs more DeepSeek API pricing ↗.
Where this comes from
Checked Oct 9, 2026. Vendor pages describe their own products; independent sources add outside views. Hands on testing is in progress.
01