Search AI Products and News

ByteDance's Seed Team recently released the impressive results of the 2025 full subject Gaokao test: the Dou Bao Seed1.6-Thinking model scored 683 in literature and 648 in science, meeting the admission threshold for Tsinghua and Peking University, and performed outstandingly in domestic and international AI model Gaokao tests. The test used the national new volume one and Shandong Province's independent proposition papers, with Dou Bao competing against five other top domestic and international AI models such as Google Gemini 2.5 Pro, DeepSeek R1, and OpenAI o3.

Recently, the "AI College Admissions Guide" service launched by QQ Browser has introduced the "College Admissions Report" feature, aiming to provide students with more scientific and reasonable admissions suggestions through intelligent methods. According to the information, students need only enter basic information such as their province, selected subjects, and scores in the "AI College Admissions Guide" on the mobile version of QQ Browser, and further specify preferences such as admission batch, region, major, and graduation intent. Within 3 to 5 minutes, they can obtain a personalized college admissions report. This report not only includes the student's personal information and strategy explanation, but also provides a detailed list of

Earth System Models are crucial tools for predicting environmental changes, helping us better prepare for the future. However, these models have extremely high computational demands, limiting their operation at fine enough resolutions. Currently, most models operate at a resolution of approximately 100 kilometers, equivalent to the size of Hawaii, making accurate predictions for specific regions difficult. Nevertheless, for practical applications such as agriculture, water resource planning, and disaster response, city-scale prediction accuracy around 10 kilometers is vital. Therefore, improving the resolution of these models is essential for better protection.

Recently, the field of artificial intelligence has been abuzz with discussion surrounding OpenAI's GPT-4o image generation model. Its exceptional performance has propelled it to the forefront in industry benchmark evaluations. According to recent social media discussions, GPT-4o's ELO score for image generation quality places it in first place alongside emerging model Reve, surpassing strong competitors such as Recraft V3, FLUX1.1[pro], and Google's Gemini2.0Flash. This achievement solidifies OpenAI's position in the generative AI field.

AI startup Luma recently announced on X that it has open-sourced a new image model pretraining technique called Inductive Moment Matching (IMM). This breakthrough technology has garnered significant attention for its efficiency and stability, considered a major advancement in the generative AI field. According to X user linqi_zhou, IMM is a novel generative paradigm capable of stable training from scratch with a single model and single objective, surpassing existing methods in both sampling efficiency and sample quality.

With the rapid development of artificial intelligence, AI tools have become increasingly popular among college students for completing reports and essays. According to CCTV News, some university teachers and experts have reported cases of academic misconduct, including fabricating research data, editing experimental images, and substituting original designs. A notice in a university assignment group drew widespread attention, stating that students submitting AI-generated essays would receive a failing grade.

Recently, researchers proposed a new method called NeuralSVG, aimed at generating vector graphics from text prompts. This innovative technology will provide artists and designers with more flexible and efficient tools to create high-quality visual content. Compared to traditional vector graphic generation methods, NeuralSVG not only generates graphics with multi-level structures but also allows users to make various dynamic adjustments during the generation process. The core of NeuralSVG lies in its implicit neural representation, achieved through a compact multi...

Douyin Group recently released the 2024 Live Streaming Governance Report, showcasing the platform's governance philosophy, methods, and achievements in the live streaming field. Over the past year, Douyin has committed to building a safe, trustworthy, and clean online live streaming environment by improving anchor management mechanisms, enhancing content identification technology, coordinating online and offline governance, and protecting key groups, resulting in significant achievements.

The latest model o3 released by OpenAI has achieved remarkable results in the ARC-AGI benchmark test, scoring an impressive 75.7% under standard computing conditions, and even reaching 87.5% under high-computation versions. This achievement has surprised the AI research community, but it still does not prove that artificial general intelligence (AGI) has been solved. The ARC-AGI benchmark test is based on the Abstract Reasoning Corpus and aims to evaluate AI systems' adaptability to new tasks.

The future development of artificial intelligence (AI) is gradually freeing itself from dependence on human data, labels, and preferences. A new AI self-learning model known as 'Socratic Learning' is being proposed, which is expected to drive AI towards genuine self-evolution. The core of this learning model is that AI improves its capabilities by interacting and questioning itself within a closed system, without external intervention from the real world. What is 'Socratic Learning'? Don't be fooled by the name; it essentially means AI plays with itself, continuously enhancing itself through dialogue and questioning.

Recently, Grammarly, a trusted AI assistant, announced the launch of a suite of new features to help businesses measure the effectiveness of communication and its impact on employee productivity and key business metrics. First, Grammarly introduced the industry's first 'Effective Communication Score', which allows corporate clients to track and assess the health of their communications. This score focuses on several key metrics, including communication accuracy, efficiency and clarity, brand compliance, and inclusive language.

Recently, the Google Quantum AI team made remarkable progress, showing that current Noisy Intermediate-Scale Quantum (NISQ) computers can complete benchmark calculations that would take classical computers years to process. This research was published in the journal Nature and utilized the random circuit sampling benchmark, which is one of the most challenging tasks currently executable by quantum computers. Image credit: This image was generated by AI, courtesy of image licensing provider Midjourney. This finding indicates that although quantum computers still have limitations.

The Google DeepMind research team has recently achieved a significant breakthrough with the development of an innovative technology called SCoRe (Self-Correction through Reinforcement Learning). This technology aims to address the long-standing challenge of large language models (LLMs) struggling to self-correct, enabling them to identify and fix errors without relying on multiple models or external checks. The core of SCoRe's technology lies in its two-phase approach. The first phase optimizes the model.

Nvidia's research team has successfully launched Llama-3.1-Minitron4B using model pruning and distillation techniques. This is a compressed version of the Llama3 model, aimed at implementing artificial intelligence on devices. The model reduces the parameter count of the original 8B model through deep and width pruning techniques while maintaining performance close to larger models. Despite a significant reduction in training data (by 40 times), the model achieved a 16% performance improvement on the MMLU benchmark.

Taobao recently announced a series of optimizations to its 'Refund Only' policy aimed at combating the improper use of this policy and enhancing merchants' autonomy in after-sales service. The new strategy will officially take effect on August 9. Firstly, Taobao will upgrade its 'Refund Only' recognition model to improve its ability to identify abnormal behaviors, such as unusually high refund requests, returning empty packages, and frequent refunds for counterfeit products. This measure aims to reject refund requests exhibiting abnormal behaviors, thereby protecting the legitimate rights and interests of both merchants and consumers.

Recently, NVIDIA launched the Minitron series of small language models, including 4B and 8B versions, significantly increasing training speed by 40 times while greatly reducing resource and data requirements, resulting in cost savings. By combining techniques of 'pruning' and 'knowledge distillation', the Minitron models maintain performance while reducing size, allowing developers to harness advanced technology for applications such as translation, sentiment analysis, and dialog AI at a lower cost. The open-source nature of the Minitron models enables more people to easily access and use them.

Claude 3 scored as high as 101 on the IQ test, outperforming GPT-4. Claude 3 has discovered a quantum algorithm that has not yet been released by physicists. Claude 3 demonstrates superior performance and can engage in scientific research. Claude 3 Opus reinvents quantum algorithms from scratch through prompts. Claude 3 excels in various tests, surpassing models like GPT-4. Claude 3 Opus showcases the evolution of self-understanding in self-portrait creation.