DeepSeek — A Chinese AI lab offering a free web/mobile chat assistant and a usage-priced API built on its open-weight DeepSeek-V4 and R1 reasoning models, released under permissive licenses that allow commercial use.
By Noahlast updated
DeepSeek screenshots & video
DeepSeek in 20 seconds
Best for: developers who want a low-cost, open-weight AI model available either via API or self-hosted download
Pricing: Chat app is free; API starts at $0.14/1M input tokens and $0.28/1M output tokens for V4 Flash, up to $0.435/$0.87 per 1M tokens for V4 Pro.
- DeepSeek's chat assistant is free at chat.deepseek.com and on iOS/Android.
- DeepSeek-V4 Flash API pricing is $0.14 per 1M input tokens and $0.28 per 1M output tokens.
- DeepSeek-V4 Pro API pricing is $0.435 per 1M input tokens and $0.87 per 1M output tokens.
- DeepSeek publishes open-weight models under a Model License permitting commercial use, plus MIT-licensed code, with 104,000+ GitHub stars on its DeepSeek-V3 repository.
- DeepSeek holds a 4.9/5 rating from 51 reviews on Product Hunt.
Limitations: Subject to Chinese data and content regulations, a concern for some enterprise and government buyers.; Hosted service has had capacity and uptime issues during demand spikes.; No official desktop app.
Top alternatives: OpenAI ChatGPT, Google Gemini, Mistral AI, Alibaba Qwen, xAI Grok — full list
What is DeepSeek?
DeepSeek is both a free consumer AI chat assistant and the developer of a family of open-weight large language models (V4-Pro, V3.2, R1) that other companies and researchers can download, fine-tune, and run themselves. The chat product at chat.deepseek.com and its iOS/Android apps are free to use, similar to ChatGPT's free tier, and support extended conversations, document upload, and coding assistance. What differentiates DeepSeek from most AI labs is that it publishes its model weights openly on GitHub and Hugging Face under a Model License permitting commercial use, alongside MIT-licensed supporting code — a stance that has made DeepSeek's models a common base for self-hosted and fine-tuned deployments. For teams that don't want to self-host, DeepSeek's own API platform charges per token, with steep discounts for cached input tokens, undercutting most closed-model API pricing from OpenAI and Anthropic on a per-token basis.
DeepSeek is an AI company whose chat assistant is free to use on the web at chat.deepseek.com and via iOS/Android apps, positioned as a low-cost, high-capability alternative to ChatGPT and other closed AI assistants. Its underlying models (V4-Pro, V3.2, R1, and earlier V3 releases) are open-weight and published on GitHub under DeepSeek's Model License, which permits commercial use, while the surrounding code is MIT-licensed; the model repository has passed 104,000 GitHub stars. Developers who want to build on DeepSeek pay usage-based API pricing: DeepSeek-V4 Flash costs $0.14 per million input tokens (cache miss) and $0.28 per million output tokens, while V4 Pro costs $0.435 per million input tokens and $0.87 per million output tokens, both with far cheaper cached-input rates.
Standout feature: Open-weight models under a Model License that permits commercial use, published alongside MIT-licensed code, letting developers self-host DeepSeek's models instead of depending solely on DeepSeek's hosted API.
Biggest weakness: DeepSeek is a Chinese company subject to Chinese data and content regulations, which has led some Western enterprises and governments to restrict or ban its use over data-residency and censorship concerns, and its hosted infrastructure has had documented uptime and capacity issues during demand spikes.
Pros
- Chat assistant is free with no subscription required for standard use.
- Model weights are open and licensed for commercial use, unlike most frontier AI labs.
- API pricing undercuts most closed-model competitors on a per-token basis, especially with cached input discounts.
- Coding and reasoning performance is competitive with closed models at a fraction of the API cost.
Cons
- Data handling and content moderation are subject to Chinese regulatory requirements, a blocker for some enterprise and government users.
- Hosted API and chat service have experienced capacity and uptime issues during high-demand periods.
- No official desktop app; access is limited to web, iOS, Android, and the API.
- Rapid model naming changes (legacy deepseek-chat and deepseek-reasoner names retired in mid-2026) can break integrations built against older model identifiers.
DeepSeek pricing
Free tier · paid from $0.14/mo
| Plan | Price | Notes |
|---|---|---|
| Chat (web/mobile) | Free | Free conversational access at chat.deepseek.com and via iOS/Android apps. |
| API — DeepSeek-V4 Flash | $0.14/mo | $0.14 per 1M input tokens (cache miss), $0.0028 per 1M input tokens (cache hit), $0.28 per 1M output tokens. |
| API — DeepSeek-V4 Pro | $0.43/mo | $0.435 per 1M input tokens (cache miss), $0.003625 per 1M input tokens (cache hit), $0.87 per 1M output tokens. |
What reviewers say
Scores belong to their platforms; we display them with attribution and never fold them into our own ratings.