News

Anysphere, the company that develops the AI programming tool Cursor, has closed a new round of funding that has seen its valuation more than triple from the start of the year to about $9 billion, showing once again that Silicon Valley AI apps startups are attracting a steady stream of funding. According to people familiar with the matter, the ...

While everyone was guessing who would come first, DeepSeek V4 or R2 or Qwen3, Qwen3 was released. In the early morning of April 29, Alibaba open-sourced a new generation of Tongyi Qianqian Qwen3 series models, covering 8 different sizes. Among them, the flagship model Qwen3 2...

What is MCP? What is MCP Model Context Protocol (MCP) is a communication protocol designed for machine learning model services to efficiently manage contextual information during model inference (e.g., session state, environment...

Endor Labs, which builds tools to scan AI-generated code for vulnerabilities, announced on April 23 that it had closed a $93 million Series B round of funding (totaling $163 million) led by DFJ Growth, with Salesforce Ventures and other...

Google Inc. made an announcement on April 17, announcing the availability of the Gemini 2.5 Flash Preview preview AI model in Google AI Studio and Vertex AI in the form of the Gemini API...

April 16, 2025 Grok has officially announced the release of the first version of Grok Studio, a revolutionary AI programming tool. As the latest product from xAI, Grok Studio not only supports code execution, but also integrates...

At 1am today, OpenAI conducted a technical live stream to release its latest model, GPT-4.1. In addition to GPT-4.1, there are two other models, GPT 4.1-Mini and GPT 4.1-Nano, which have been developed in the areas of multimodal processing, code capability, instruction adherence, and into...

Stanford University's HAI Research Center has released the 2025 AI Index Report. The report covers a comprehensive analysis of many aspects from AI hardware and technology performance to responsible AI application, development, and policy governance, and is one of the most in-depth and authoritative reports analyzing AI in the world. The official website ...

Look at the market: fintech development enters the golden age, AI market scale is rapidly rising The global artificial intelligence in fintech market size is expected to climb to $14.41 billion in 2024, benefiting from the deep application of intelligent risk control, algorithmic trading and customer service, and in the next five years it will be 27.1% in...

Runway, a company specializing in video generation, officially launched their new generation of model series -- Gen-4, which is a story that the director has been brewing for more than ten years but has never been able to realize with AI. Prior to this, to create this video with AI, a series of problems need to be overcome: the corner...

This week, Midjourney is coming back strong with the new V7. The internal model has been trained, and the scoring system is now turned on, entering the final fine-tuning stage. Netizens have already released the birth charts, the effect is stunning, and the picture quality and fineness are pulled full circle. GPT-4o heat is not yet over, Midjourney...

No launch event, no overwhelming publicity, on March 24th DeepSeek version V3-0324 went live quietly. https://huggingface.co/deepseek-ai/DeepSeek-V3-032...

Blackwell has not yet been delivered on a large scale, and NVIDIA has already laid out two generations of successors. On Tuesday, March 18 local time, NVIDIA CEO Jen-Hsun Huang delivered a keynote speech at the GTC25 conference, announcing the 2026-2027 data center GPU roadmap, Rubin and Ru...

March 16, Baidu officially released Wenxin Big Model 4.5 and Wenxin Big Model X1, which can be used free of charge on the official website of Wenxin Yiyin. According to reports, Wenxin Big Model 4.5 is Baidu's first native multimodal big model, and its multimodal comprehension, text and logical reasoning capabilities have been significantly improved in a number of measurement...

In recent times, the inference model DeepSeek-R1 has arguably been the number one topic in AI. Those who have used it know that the model outputs a piece of thought chain content before outputting the final answer. Doing so improves the accuracy of the final answer. Today's post will take you through the chain of thought...

The Gemma series of macromodels is a series of lightweight macromodels open-sourced by Google. Just a few moments ago (March 12, 2025), Google open-sourced the third generation of the Gemma series of macromodels, which contains a total of four different parameter scale versions.The third generation of the Gemma 3 series is multi...

At present, the application of big models in the field of government affairs has become an important hand of the government to improve the level of service, DeepSeek deployment and application in the field of government affairs around China is advancing at an unprecedented speed.DeepSeek series of models, by virtue of its advantages in terms of cost and performance, in the field of government services, public...

In the early morning of March 6, another sleepless night in the tech circle after DeepSeek, everyone was screened by a product called Manus.The AI circle was boiling, and the AI intelligent body plate soared.The release of Manus made Chinese AI technology shock the world once again. According to its team...

With the rapid development of artificial intelligence technology, large-scale language modeling (LLM) has become a key force driving this field forward. In order to better master and utilize LLM technology, it is particularly important to understand its core parameters. In this paper, we will take an in-depth look at three key parameters in large-scale language modeling: the Toke...

In today's era of rapid development of artificial intelligence, DeepSeek, as a leading AI model, has become the first choice of many enterprises due to its powerful functions and wide range of application areas. On the one hand, R1, V3 and other versions of the model, with the label of "performance comparable to GPT-4, cost only 10%", have pushed ...