Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Find out more
As of a few days ago, only the most capable of nerds (I say this as one) have ever heard of it Deepseeka Chinese subsidiary of equally evocatively named AI High flyer capital managementa quantitative (or quantitative) analysis company that initially launched in 2015.
However, over the past few days, it has probably been the most talked about company in Silicon Valley. This is mainly thanks to the release of Deepseek R1, a novel huge language model that performs “reasoning” similar to Openai’s current best-available O1 model – it takes many seconds or minutes to answer arduous questions and solve elaborate problems because it reflects its own analysis analysis in a step-by-step or “chain of thought” way.
Not only that, but Deepseek R1 rated as high or higher than Openai’s O1 on various external benchmarks (tests to measure AI performance when answering questions on various topics trained at a fraction of the cost (reportedly around $5 million) with significantly fewer graphics processing units (GPUs) under a strict U.S. embargo on Openai’s home turf.
But unlike O1, which is only available to paying Chatgpt subscribers of the Plus tier ($20/month) and more high-priced tiers (like PRO at $200/month), Deepseek R1 has been released as a fully open source model, which also explains why also explains why explains why explains why explains why explains why explains why explains He quickly picked up the charts AI Code Shaving Community Community Hugging Face most downloaded and active models.
Furthermore, thanks to the fact that it is fully open source, people have already adapted and trained many many variations of the model for various task-specific purposes, such as making it tiny enough to run on a mobile device or combining other open source models. Even if you want to exploit it for development purposes, Deepseek API costs are over 90% cheaper than the equivalent O1 model from Openai.
Most impressively, you don’t even have to be a software engineer to exploit it: Deepseek has Free website AND Mobile application even for us users With an R1 powered chatbot interface very similar to Chatgpt Openai. Except, once again, Openai’s Deepseek Ondeckut or “Mogged” plugs in this powerful reasoning model with network search – something Openai has yet to do (network search is only available in the less powerful GPT family of models).
Open and closed irony
There is some rather delicious, and perhaps disturbing, irony in Openai’s founding goals of democratizing AI for the masses. How Jim Fan, senior research manager at Nvidia, put it on x: “We live on a schedule in which a non-US company keeps the original Openai mission alive – truly open, frontier research that empowers everything. This doesn’t make sense. The most fun outcome is the most likely. “
Or as X User @SuspendedRobot Embed thisreferring to reports It appears that Deepseek has been trained In question results and other data generated by Chatgpt: “Openai stole from all over the internet to make himself richer, Deepseek stole from them and returned it to the masses for free
But Fan isn’t the only one who has sat back and taken note of Deepseek’s success. The open source availability of Deepseek R1, its high performance, and the fact that it seemingly “came out of nowhere” to challenge the former leader in generative AI has sent shockwaves through Silicon Valley and far, based on my conversations and readings with various engineers, thinkers and leaders . If not “everyone” is freaking out about this, as my hyperbolic headline suggests, it’s certainly the talk of the town in tech and business circles.
AND A message published to the blindapp for sharing anonymous rumors in Silicon Valley, rounds suggest that Meta is in crisis due to Deepseek’s success due to how quickly it has surpassed Meta’s own efforts to become the king of open source AI with LLAMA models.
“This changes the whole game.”
X User @tphuang wrote strongly: “Deepseek has established artificial intelligence beyond the very highest levels. Lightbulb moment for me in the 1st photo. R1 is so much cheaper than the US labor cost that many jobs will be automated over the next 5 years,” later I notice Why R1 Deepseek is more tempting to users than even O1 Openai:
“3 huge problems with the O1:
1) Too slow
2) Too expensive
3) Lack of end user control/reliance on Openai
R1 solves them all. The company can buy its own NVIDIA GPU, run these models. I don’t have to worry about additional costs or slow/unresponsive Openai servers”
@tphaung also bet compelling analogy as a question: “Will Deepseek be an LLM what Android has become to the OS world?”
Internet entrepreneur Arnaud Bertrand did not mention the surprising implications of Deepseek’s success, Writing on x: “It’s no exaggeration how profoundly this changes the entire game. And not only in relation to artificial intelligence, it is also a huge indictment of China’s misguided attempt to stop technological development, without which Deepseek may not have been possible (as the saying goes, necessity is the mother of invention).”
The issue of censorship
However, others sounded cautious notes about Deepseek’s rapid growth, arguing that as a startup operated from China, it was necessarily subject to that country’s regulations and content censorship requirements.
Indeed, my own exploit of Deepeek on the iOS app here in the US has discovered don’t answer questions about Tiananmen Squaresite of student protests and student uprising in 1989 causing at least 200, possibly thousands of deathsearning this nickname “Tiananmen Square Massacre”In Western media.
Ben Hyrak, former Apple Human Interface designer and co-founder of the AI Analytics platform AI Dawn, published on X How Asking about this topic sent Deepseek R1 into a circular loop.
As a member of the press itself, I obviously take freedom of speech and expression very seriously, and this is perhaps one of the most fundamental, enduring causes of I Champion.
What cannot be noted, however, is that Openai’s models and products, including Chatgpt, also refuse to answer a whole range of questions about even innocent content – particularly regarding human sexuality and erotic/adult NSFW topics.
Of course, this is not an apples-to-app comparison. There will be some for whom resistance to relying on foreign technology makes them skeptical of Deepseek’s ultimate value and utility. But you can’t deny its efficiency and low cost.
And at a time when 16.5% of all US goods are imported through ChinaI find it arduous to warn against using Deepseek R1 based on censorship concerns or security threats – especially when the model code is freely downloaded, taken offline, exploit device on device in secure environments and adapt at will.
I definitely detected an existential crisis about the “fall of the West” and the “Rise of China”, motivating some animated discussions about Deepseek, and others have already linked it to HHOw Us Users Joined Xiaohongshu (aka “Little Red Book”) App When Tiktok was briefly banned in the country, only to be surprised by the quality of life in China depicted in the videos shared there. The arrival of Deepseek R1 occurs in this narrative context – in which China appears (and by many indicators it clearly does), while the US appears (and in many indicators it does as well).
The first, but almost last, Chinese AI model that shook the world
Nor will this be the last Chinese AI model to threaten the dominance of Silicon Valley giants – even if, like Openai, they raise more money than ever for their ambitions to develop artificial general intelligence (AGI), programs that at most outperform humans, which outperform people’s economically valuable work.
Just yesterday, another Chinese model from Tiktok Matter Company Bytedance – called Doubao-1.5-Pro -It was released with the unjustified GPT-4O model in Performance, but on third party references but Again, about 1/50 the cost.
Chinese models have become so good, so swift, even for non-tech people: Economist The magazine has just had a slice of Deepseek’s success and other Chinese AI efforts and political commentator Matt Bruenig posted on x that: “I have been using Gemini, Chatgpt, and Claude extensively to summarize NLRB documents for almost a year. Deepseek is better than everyone. The Chatbot version is free. The price to exploit the API is 99.5% below the Openai API price. [shrug emoji]”
How does Openai respond?
Nicki Wonder co-founder and CEO Sam Altman Today she said the company He brought his future family the Second Reasoning Model, O3, to chatgpt even for free users. Openai continues to carve its own path with more proprietary and advanced models – setting the industry standard.
But the question arises: with Deepsek, Bytedance and other Chinese AI companies hitting their heels, how long can OpenAi stay at the forefront of creating and releasing novel cutting-edge AI models? And if and when he falls, how strenuous and how swift will he fall?
Openai, however, has another historical precedent. If Deepseek and China’s AI models actually become LLM like Google’s Open Source Android to Mobile – taking on the lions in the market for a while – you just have to see Apple iPhone go with a closed, proprietary, all-in. The home approach has managed to carve out a high-end market and from here expand downwardsparticularly in the US, so much so that it now owns almost 60% of the domestic smartphone market.
Still, for all those who spend gigantic bucks to exploit AI models from leading labs, Deepseek shows that the same capabilities can be available for much cheaper and with much more control. And in a corporate setting, that may be enough to win the ball.
