We have a breakthrough new player on the artificial intelligence field: DeepSeek is an AI assistant developed by a Chinese company called DeepSeek. Thanks to social media, DeepSeek has been breaking the internet for the last few days.
Earlier in January, DeepSeek released its AI model, DeepSeek (R1), which competes with leading models like OpenAI's ChatGPT o1. What sets DeepSeek apart is its ability to develop high-performing AI models at a fraction of the cost.
It has a user-friendly design. It's built to assist with various tasks, from answering questions to generating content, like ChatGPT or Google's Gemini. But unlike the American AI giants, which usually have free versions but impose fees to access their higher-operating AI engines and gain more queries, DeepSeek is all free to use.
It also quickly launched an AI image generator this week called Janus-Pro, which aims to take on Dall-E 3, Stable Diffusion and Leonardo in the US.
So what makes DeepSeek different, how does it work and why is it gaining so much attention?
The founding of DeepSeek
Founded in 2023 by a hedge fund manager, Liang Wenfeng, the company is headquartered in Hangzhou, China, and specializes in developing open-source large language models.
Because it is an open-source platform, developers can customize it to their needs. Little known before January, the AI assistant launch has fueled optimism for AI innovation, challenging the dominance of US tech giants that rely on massive investments in chips, data centers and energy.
How DeepSeek works
DeepSeek operates as a conversational AI, meaning it can understand and respond to natural language inputs. You can ask it a simple question, request help with a project, assist with research, draft emails and solve reasoning problems using DeepThink.
DeepSeek offers two LLMs: DeepSeek-V3 and DeepThink (R1). DeepSeek-V3 works like the standard ChatGPT model, providing fast responses, generating text, rewriting emails and summarizing documents. DeepThink (R1) provides an alternative to OpenAI's ChatGPT o1 model, which requires a subscription, but both DeepSeek models are free to use.
They can be accessed via web browsers and mobile apps on iOS and Android devices. In fact, by late January 2025, the DeepSeek app became the most downloaded free app on both Apple's iOS App Store and Google's Play Store in the US and dozens of countries globally.
DeepSeek uses advanced machine learning models to process information and generate responses, making it capable of handling various tasks.
Also setting it apart from other AI tools, the DeepThink (R1) model shows you its exact "thought process" and the time it took to get the answer before giving you a detailed reply.
Perplexity now also offers reasoning with R1, DeepSeek's model hosted in the US, along with its previous option for OpenAI's o1 leading model.
Perplexity now offers DeepSeek R1.
Self-censoring, data privacy and other concerns
Trust is key to AI adoption, and DeepSeek could face pushback in Western markets due to data privacy, censorship and transparency concerns. Similar to the scrutiny that led to TikTok bans, worries about data storage in China and potential government access raise red flags.
There's also fear that AI models like DeepSeek could spread misinformation, reinforce authoritarian narratives and shape public discourse to benefit certain interests.
For example, when asked about sensitive topics like the Tiananmen Square massacre in 1989, the status of Taiwan or other politically charged issues, DeepSeek initially provided accurate responses but self-censored within seconds, replacing them with a generic message: "Sorry, that's beyond my current scope. Let's talk about something else."
Sometimes, it skipped the initial full response entirely and defaulted to that answer. Another common deflection was: "Let's chat about math, coding and logic problems instead!"
DeepSeek's deflection when asked about controversial topics that are censored in China.
US-based AI companies have had their fair share of controversy regarding hallucinations, telling people to eat rocks and rightfully refusing to make racist jokes. The problem with DeepSeek's censorship is that it will make jokes about US presidents Joe Biden and Donald Trump, but it won't dare to add Chinese President Xi Jinping to the mix.
DeepSeek tells a joke about US Presidents Biden and Trump, but refuses to tell a joke about Chinese President Xi Jinping.
DeepSeek didn't immediately respond to a request for comment about its apparent censorship of certain topics and individuals.
Innovative technology and cost-efficiency
DeepSeek is making headlines for its performance, which matches or even surpasses top AI models. Its R1 model outperforms OpenAI's o1-mini on multiple benchmarks, and research from Artificial Analysis ranks it ahead of models from Google, Meta and Anthropic in overall quality.
Aside from benchmarking results that often change as AI models upgrade, the surprisingly low cost is turning heads. The company claims to have built its AI models using far less computing power, which would mean significantly lower expenses. However, these figures haven't been independently verified.
DeepSeek-R1 was allegedly created with an estimated budget of $5.5 million, significantly less than the $100 million reportedly spent on OpenAI's GPT-4. This cost efficiency is achieved through less advanced Nvidia H800 chips and innovative training methodologies that optimize resources without compromising performance.
However, some experts and analysts in the tech industry remain skeptical about whether the cost savings are as dramatic as DeepSeek states, suggesting that the company owns 50,000 Nvidia H100 chips that it can't talk about due to US export controls. DeepSeek didn't immediately respond to a request for comment.
Market disruption and global impact
These claims still had a massive pearl-clutching effect on the stock market. Forbes reported that Nvidia's market value "fell by about $590 billion Monday, rose by roughly $260 billion Tuesday and dropped $160 billion Wednesday morning." Other tech giants, like Oracle, Microsoft, Alphabet (Google's parent company) and ASML (a Dutch chip equipment maker) also faced notable losses.
DeepSeek's rapid rise has disrupted the global AI market, challenging the traditional perception that advanced AI development requires enormous financial resources. Marc Andreessen, an influential Silicon Valley venture capitalist, compared it to a "Sputnik moment" in AI.
While DeepSeek has earned praise for its innovations, it has also faced challenges. The company experienced cyberattacks, prompting temporary restrictions on user registrations.
While Trump called DeepSeek's success a "wakeup call" for the US AI industry, OpenAI told the Financial Times that it found evidence DeepSeek may have used its AI models for training, violating OpenAI's terms of service. White House AI adviser David Sacks confirmed this concern on Fox News, stating there is strong evidence DeepSeek extracted knowledge from OpenAI's models using "distillation." It's a technique where a smaller model ("student") learns to mimic a larger model ("teacher"), replicating its performance with less computing power.
Despite the controversies, DeepSeek has committed to its open-source philosophy and proved that groundbreaking technology doesn't always require massive budgets. As we have seen in the last few days, its low-cost approach challenged major players like OpenAI and may push companies like Nvidia to adapt. This opens opportunities for innovation in the AI sphere, particularly in its infrastructure.