Latest news with #人工智能
South China Morning Post
a day ago
- Business
- South China Morning Post
China's DeepSeek closes in on US rival OpenAI, surpasses Alibaba with upgraded model
Chinese artificial intelligence (AI) start-up DeepSeek said R1-0528, the first significant upgrade to its R1 reasoning model that debuted in January, matched the performance of top global competitors, including OpenAI and Google. Advertisement In a statement released late on Thursday, DeepSeek highlighted improvements in the new model's reasoning and creative writing capabilities, making it more adept at crafting argumentative essays, fiction and prose in styles that closely mimic human authors. Coding capabilities have also been enhanced. The company said the latest version achieved a 50 per cent reduction in 'hallucinations' – instances where AI generates misleading information with little factual basis. These upgrades were achieved by investing additional computing resources in the post-training stage, when developers make final adjustments and enhancements to the model after the main training process, the company said. Post-training usually focuses on boosting efficiency and enhancing content safety and accuracy. 'The updated R1 model excelled among domestic AI models in a range of benchmark tests, including maths, coding and general logic, and matched up to global top models such as [OpenAI's] O3 and [Google's] Gemini2.5-Pro,' DeepSeek said. Benchmark results cited by DeepSeek shows that R1-0528 outperforms Alibaba's Qwen3 AI model. Photo: Shutterstock The update comes after the original R1 model was dethroned in late April by Alibaba Group Holding's flagship model, Qwen3, in the LiveBench rankings for leading open-source AI systems. The shift underscores the heated competition among Chinese tech players in advancing AI capabilities.
South China Morning Post
3 days ago
- Business
- South China Morning Post
DeepSeek quietly updates R1 AI model amid anticipation for next-gen tech
Chinese artificial intelligence (AI) start-up DeepSeek quietly released a new version of its R1 reasoning model on Wednesday, marking its first revision since its high-profile debut in January. The Hangzhou-based company said it had 'completed a minor update to the R1 model', which is now available on the website for its namesake chatbot, as well as its mobile apps, according to a notice posted in a company-run WeChat group chat. DeepSeek did not disclose details of the changes in the update, dubbed R1-0528, which is now live on the open-source AI platform Hugging Face. DeepSeek did not immediately respond to a request for comment. The company last updated its foundational large language model V3 in March, touting improvements in coding and writing in the V3-0324 release on Hugging Face.
South China Morning Post
3 days ago
- Business
- South China Morning Post
Smartphone maker Honor joins robotics race after pledging US$10 billion AI investment
Chinese smartphone maker Honor has joined the country's heated race to develop robots, as it aims to reposition itself as an artificial intelligence (AI) player amid heightened competition in the Android handset market. Honor said on Wednesday that it would develop its own robots, and that it had already helped Chinese start-up Unitree Robotics break the record for running speed by a humanoid robot. The Shenzhen-based firm, a spin-off from telecommunications giant Huawei Technologies, said that its robotics efforts would include working with partners 'to enable more possibilities'. Honor's intended foray into China's crowded robotics space comes after its newly appointed CEO James Li Jian announced a high-profile AI initiative earlier this year that he called the 'Honor Alpha plan'. The plan will see the firm invest US$10 billion over the next five years to transform itself from a smartphone maker into 'an ecosystem company' focused on AI devices, Li said ahead of the MWC Barcelona trade show in Spain in March. Honor used its proprietary AI algorithm to train a robot from Chinese start-up Unitree. The machine achieved a peak running speed of 4-metres per second, breaking the record for humanoid robots, the company said on Wednesday at a launch event in Shenzhen for its new Honor 400 series handsets.
Bloomberg
4 days ago
- Business
- Bloomberg
Kuaishou Sees Higher Spending on AI Development Denting Margins
Chinese social platform Kuaishou Technology expects to spend more on artificial intelligence development as it tries to dispel concerns about lackluster monetization of its latest AI tool. Overall losses for Kling AI will continue to narrow as revenue grows, while fixed costs like training and labor will be more spread out, according to Chief Financial Officer Jin Bing.
South China Morning Post
21-05-2025
- Business
- South China Morning Post
‘What if AI had eyes': meet Chinese firm Xreal, Google's partner in AR glasses project
Chinese augmented reality (AR) firm Xreal is back in the spotlight via a partnership with Google to create the first eyewear powered by the US tech giant's Android XR operating system, intensifying competition in the nascent market for spectacles supported by artificial intelligence (AI) technology. The start-up 's venture with Google, dubbed Project Aura, was among the key announcements made on Tuesday at the opening of the US internet search company's annual I/O developer conference, which concluded on Wednesday. Project Aura – with the tagline 'what if AI had eyes' – features a pair of lightweight, tethered Xreal AR glasses powered by Qualcomm 's Snapdragon processor. The smart glasses will offer a dual-mode augmented reality experience; optical see-through, which is a direct overlay of digital content onto a real-world view; and virtual see-through, which blends camera-captured images with digital elements, according to Google and Xreal. Without providing more details, Xreal founder and chief executive Xu Chi said the new AR glasses are expected to hit the consumer market later in 2025 or early next year, according to his interview with Bloomberg.
