Skip to main content
D

Deepseek

4.4(206 reviews)

4 comparisons available

About Deepseek

DeepSeek is a Chinese AI company founded in 2023 as a subsidiary of quantitative hedge fund High-Flyer, headquartered in Hangzhou, China. It rose to global prominence in January 2025 when DeepSeek R1 — a reasoning model comparable to OpenAI's o1 — was released as open source and became the top-downloaded app on the App Store globally. DeepSeek's models are notable for being trained at a fraction of the cost of comparable Western models: DeepSeek V3 reportedly cost approximately $5.6 million to train, versus hundreds of millions for comparable GPT-4 class models, achieved through novel efficiency techniques including Mixture of Experts (MoE) architecture and FP8 training. DeepSeek R1 and V3 are fully open source (MIT license), enabling self-hosting and commercial use. The models benchmark competitively with GPT-4o and Claude 3.5 Sonnet on coding, math, and reasoning tasks. DeepSeek's release triggered significant market reaction — Nvidia's stock dropped 17% in a single day as investors questioned GPU demand assumptions. The DeepSeek app is free with no usage caps, though the app raised privacy concerns due to data storage in China under Chinese law.

R1 became #1 App Store download globally in Jan 2025Trained V3 for ~$5.6M vs hundreds of millions for GPT-4Fully open source (MIT license) — self-hostableMixture-of-Experts architecture enabling massive efficiency gains

Frequently Asked Questions

Is DeepSeek safe to use?

DeepSeek's app raises legitimate privacy concerns: the company is Chinese, data is stored on servers in China, and Chinese law requires companies to share data with the government on request. Several governments (Italy, Australia, the US Navy) have restricted its use on official devices. For personal use and non-sensitive queries, the privacy risk is similar to other free AI apps. For enterprise or government use with sensitive data, avoid DeepSeek or use the open-source model self-hosted on your own infrastructure.

How good is DeepSeek compared to ChatGPT?

DeepSeek R1 benchmarks very competitively with OpenAI's o1 on math, coding, and reasoning tasks — sometimes surpassing it. DeepSeek V3 is comparable to GPT-4o. The key differences: DeepSeek is free with no usage caps (via the app), open source, and extremely cost-efficient via API. ChatGPT has a larger ecosystem, better multimodal capabilities (voice, image generation), stronger enterprise trust, and US data residency. For pure reasoning quality at zero cost, DeepSeek R1 is exceptional.

Can I use DeepSeek for free?

Yes — the DeepSeek app (deepseek.com and mobile apps) is free with no usage limits on DeepSeek V3 and R1 models. DeepSeek API pricing is very competitive: V3 input tokens cost $0.27/million (cache hit) and $0.07/million (cache miss for long contexts) — significantly cheaper than GPT-4o. The open-source models can be downloaded from Hugging Face and self-hosted at no cost beyond compute.