China's DeepSeek releases update to AI model that sent US shares tumbling earlier this year
Chinese artificial intelligence startup DeepSeek released an update to its R1 reasoning model in the early hours of Thursday, stepping up competition with US rivals such as OpenAI.
DeepSeek launched R1-0528 on developer platform Hugging Face, but has yet to make an official public announcement. It did not publish a description of the model or comparisons.
But the LiveCodeBench leaderboard, a benchmark developed by researchers from UC Berkeley, MIT, and Cornell, ranked DeepSeek's updated R1 reasoning model just slightly behind OpenAI's o4 mini and o3 reasoning models on code generation and ahead of xAI's Grok 3 mini and Alibaba's Qwen 3.
Bloomberg earlier reported the update on Wedneday. It said that a DeepSeek representative had told a WeChat group that it had completed what it described as a 'minor trial upgrade' and that users could start testing it.
DeepSeek earlier this year upended beliefs that US export controls were holding back China's AI advancements after the startup released AI models that were on a par with or better than industry-leading models in the United States at a fraction of the cost.
The launch of R1 in January sent tech shares outside China plummeting in January and challenged the view that scaling AI requires vast computing power and investment. Since R1's release, Chinese tech giants like Alibaba and Tencent have released models claiming to surpass DeepSeek's.
Google's Gemini has introduced discounted tiers of access while OpenAI cut prices and released an o3 mini model that relies on less computing power.
The company is still widely expected to release R2, a successor to R1. Reuters reported in March, citing sources, that R2's release was initially planned for May. DeepSeek also released an upgrade to its V3 large language model in March.
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Bloomberg
19 minutes ago
- Bloomberg
Japan's Hostile Takeover Boom Raises Risks for Megabanks
Hi, it's Taro Fuse in Tokyo, looking at how Japan's megabanks are responding to the rising level of hostile bids in the country. Also today, Ascension Health is nearing a $3.9 billion deal for surgery centers and the 49ers empire now includes Scotland. Today's top stories Japanese megabanks play a powerful role in the country's business landscape—forming deep roots with their corporate customers over the decades through lending relationships, advisory work and even owning stakes.
TechCrunch
19 minutes ago
- TechCrunch
It's not your imagination: AI is speeding up the pace of change
If the adoption of AI feels different from any tech revolution you may have experienced before — mobile, social, cloud computing — it actually is. Venture capitalist Mary Meeker just dropped a 340-page slideshow report — which used the word 'unprecedented' on 51 of those pages — to describe the speed at which AI is being developed, adopted, spent on, and used, backed up with chart after chart. 'The pace and scope of change related to the artificial intelligence technology evolution is indeed unprecedented, as supported by the data,' she writes in the report, called 'Trends — Artificial Intelligence.' There's a certain poetic history to this person writing this kind of report. Meeker is the founder and general partner at VC firm Bond, and was once known as Queen of the Internet for her previous annual Internet Trends reports. Before founding Bond, she ran Kleiner Perkins' growth practice, from 2010-2019, where she backed companies like Facebook, Spotify, Ring, and Block (then Square). She hasn't released a trends report since 2019. But she dusted off her skills to document, in laser detail, how AI adoption has outpaced any other tech in human history. ChatGPT reaching 800 million users in 17 months: unprecedented. The number of companies and the rate at which so many others are hitting high annual recurring revenue rates: also unprecedented. The speed at which costs of usage are dropping: unprecedented. While the costs of training a model (also unprecedented) is up to $1 billion, inference costs — for example, those paying to use the tech — has already dropped 99% over two years, when calculating cost per 1 million tokens, she writes, citing research from Stanford. Techcrunch event Save now through June 4 for TechCrunch Sessions: AI Save $300 on your ticket to TC Sessions: AI—and get 50% off a second. Hear from leaders at OpenAI, Anthropic, Khosla Ventures, and more during a full day of expert insights, hands-on workshops, and high-impact networking. These low-rate deals disappear when the doors open on June 5. Exhibit at TechCrunch Sessions: AI Secure your spot at TC Sessions: AI and show 1,200+ decision-makers what you've built — without the big spend. Available through May 9 or while tables last. Berkeley, CA | REGISTER NOW The pace at which competitors are matching each other's features, at a fraction of the cost, including open source options, particularly Chinese models: unprecedented. For example, she points out that Nvidia's 2024 Blackwell GPU uses 105,000x less energy per token than the company's 2014 Kepler GPU predecessor. Meanwhile chips from Google, like its TPU (Tensor Processing Unit) and Amazon's Trainium are being developed at scale for their clouds — that's moving quickly, too. 'These aren't side projects — they're foundational bets,' she writes. The one area where AI hasn't outpaced every other tech revolution is in financial returns. While VCs are pouring money on the AI fire as fast as they can, AI companies and cloud service providers are also burning though cash. AI requires massive investments in infrastructure. That's good for consumers and enterprises, the beneficiaries of fast improvements while competition lowers costs, Meeker points out. But the jury is still out over which of the current crop of companies will become long-term, profitable, next-generation tech giants. 'Only time will tell which side of the money-making equation the current AI aspirants will land,' she writes. As of for the rest of us: Just hold onto your hats.
Yahoo
20 minutes ago
- Yahoo
International Chauffeur Service Webus Plans $300M Raise for XRP Strategic Reserve
Webus International, a China-based provider of customizable car and touring services for travelers worldwide, said it plans to raise as much as $300 million through non-equity financing to establish an XRP XRP reserve supporting its global chauffeur payment network. The reserve is part of a broader push to integrate XRP's cross-border settlement capabilities into Webus's ecosystem, including on-chain booking records and a Web3-based loyalty program, the company said. The initiative aims to streamline international payments while preserving equity value by relying on loans, credit lines and shareholder guarantees rather than issuing new shares, CEO Nan Zheng said in the statement. The announcement follows Saudi Arabia-based VivoPower International's plans to build a $121 million XRP treasury, indicative of growing interest from companies in the fourth-largest token by market cap. The financing plan is non-binding and subject to final agreements and due diligence. Webus added it was renewing a partnership with Tongcheng Travel Holdings, one of China's largest online travel agencies, to extend their 'Wetour x Tongcheng' charter lines — stating they would aim to use the XRP Ledger to settle cross-border rides and driver payouts. Sign in to access your portfolio
