This article explains why most LLM‑based agents only perform pseudo‑planning through prompts or hard‑coded loops, outlines when to rely on prompt‑driven versus program‑driven planning, compares popular frameworks such as ReAct, MRKL, BabyAGI and AutoGPT, and clarifies what true autonomous planning would require.
This article distinguishes informatization from digitalization by comparing their core focus, goals, thinking patterns, technologies, value creation, and evolutionary relationship, using examples like expense reimbursement and procurement to illustrate how digitalization leverages data to transform business models beyond mere process efficiency.
This article explains the limitations of basic Retrieval‑Augmented Generation (RAG), introduces Rerank technology as a two‑step refinement process, compares dual‑encoder and cross‑encoder methods, and reviews popular Rerank models to help developers build more precise AI‑driven retrieval systems.
This article analyzes the rStar2‑Agent paper, revealing how Agentic Reinforcement Learning, the GRPO‑RoC algorithm, a high‑throughput code‑execution service, and a three‑stage training recipe let a modest 14‑billion‑parameter model surpass much larger LLMs on challenging math benchmarks.
A recent ISE survey reveals that AI is rapidly displacing entry‑level roles in the UK tech sector, causing a steep decline in graduate hiring and raising concerns about a long‑term talent shortage for future mid‑level professionals.
Anthropic's newly released Claude Haiku 4.5 offers a small, fast, cost‑effective AI model whose benchmark results rival Sonnet 4 and even compete with leading models like Gemini 2.5 and GPT‑5, making it ideal for multi‑agent applications and developers seeking high performance at low price.
Oracle, once a second‑tier cloud provider, is accelerating its growth with a 25% annual rate, expanding AI‑focused infrastructure, massive GPU clusters, and an AI Data Platform that integrates private data, positioning itself to challenge the dominant hyperscale clouds.
The article explains that the greatest risk in AI is talent scarcity, argues that companies should skip hiring races and instead transform current engineers, analysts, and architects into AI developers, and offers practical steps for upskilling staff using familiar technologies while minimizing risk.
This article introduces several noteworthy open‑source AI projects—including Google’s Gemini CLI, the AI‑Engineering‑Hub learning repository, and OpenAI’s GPT‑5 coding examples—providing URLs, key features, and visual previews to help developers quickly explore and adopt cutting‑edge AI tools.
This article explains the definitions, technical implementations, functions, limitations, and collaborative workflow of short‑term and long‑term memory in large‑language‑model agents, detailing context windows, attention mechanisms, vector storage, retrieval strategies, and future research directions for building personalized, continuously learning AI agents.
The article explains the differences between AI Agents and AI Workflows, compares their characteristics, introduces the hybrid Agentic Workflow concept, and offers practical recommendations for building enhanced LLM applications using simple prompts or advanced frameworks.
The article surveys expert predictions for the 2025 Nobel Prizes, highlighting AI‑powered breakthroughs in physiology or medicine, physics and chemistry—including protein‑structure prediction, quantum information, next‑generation DNA sequencing, microbiome research, and novel weight‑loss drugs—while outlining key candidates and timelines.
This article explains how Xiaomi’s XiaoAI assistant integrates large language models for intent distribution, vertical intent understanding, and response generation, detailing the architecture, challenges such as knowledge requirements and sub‑200 ms latency, and the shift from prompt engineering to model fine‑tuning that boosted user retention by 10% and query satisfaction by 8%.
The article analyzes the Sloan Research, Gairdner, and Lasker awards—highlighting their recent winners, the overlap with Nobel laureates, and the growing influence of AI in science—to anticipate possible directions for the upcoming Nobel Prize announcements.
Modern warfare has evolved into an information‑centric, intelligent contest where mathematical models—ranging from classic Lanchester combat equations and resource allocation optimizations to game‑theoretic and Bayesian approaches—enable commanders to predict battlefield dynamics, allocate firepower, assess situations, and plan optimal paths and strategies.
The open‑source UI‑UG‑7B multimodal model from Alipay combines UI understanding and generation in a single framework, delivering state‑of‑the‑art performance across referring, grounding, captioning, and code generation tasks while dramatically speeding up UI creation for developers.
Li Kaifu argues that open‑source large‑model ecosystems are essential for China to close the AI gap with the United States, highlighting DeepSeek’s impact, shifting scaling laws, and the emerging role of AI‑to‑AI teaching as the next development frontier.
This article outlines the lifecycle of large language models and presents LLMOps best practices—including data management, model development, deployment, monitoring, prompt engineering, and security—to help engineers build, scale, and maintain production-ready LLM applications.
This curated list presents the most influential AI books, covering foundational theory, machine learning, deep learning, reinforcement learning, computer vision, and AI ethics, with editorial insights and author biographies to guide readers through the evolving landscape of artificial intelligence.
AI ethicsReinforcement Learningartificial intelligence
This article explains Retrieval‑Augmented Generation (RAG), an AI framework that combines traditional information retrieval with large language models, covering its core workflow—from knowledge preparation, data cleaning, and metadata extraction to query preprocessing, vector retrieval, reranking, information integration, and final LLM generation, while also reviewing common embedding models and vector databases.
Preparing for large-model interview? This guide reveals why interviewers probe seemingly minor components—positional encoding, residual connections, layer normalization, and feed-forward networks—explains each technique's purpose, variants, and how to answer confidently, plus practical tips and a learning roadmap to boost your chances.
Late at night on the 23rd, Lin Junyang of Tongyi Lab announced six AI model releases—including a safety‑audit guard, a personal travel planner, a real‑time multilingual translator, upgraded coding models, a powerful vision‑language model, and the flagship Qwen3‑Max—each detailed with capabilities, highlights, and direct download links.
Large Language Modelsartificial intelligencecoding
Stanford’s CS231n Spring 2025 course, led by Fei‑Fei Li and a team of leading AI researchers, is now fully available online with video lectures, detailed syllabus, instructor bios, and prerequisite guidelines, offering a comprehensive deep‑learning curriculum for computer‑vision enthusiasts.
In his 2025 Cloud Conference keynote, Alibaba Cloud CEO Wu Yongming outlines a three‑stage roadmap from AGI to superintelligent AI, describes large models as the next operating system, and explains the massive infrastructure and open‑source strategy needed to power the future AI era.
Oracle announced that Safra Catz will share the CEO role with Clay Magouyrk and Mike Sicilia, a move aimed at strengthening its cloud infrastructure and AI capabilities, while highlighting record financial performance and outlining the strategic importance of OCI, AI-driven growth, and the company's involvement in rebuilding TikTok's algorithm for U.S. security.
The article examines how large‑model AI is reshaping R&D by advancing intelligent development tools—from early code‑completion assistants like GitHub Copilot to expansive code‑generation platforms—highlighting their evolution, enterprise adoption, practical developer experiences, and future workflow transformations.
DeepSeek released the V3.1‑Terminus model, fixing the notorious “extreme” character bug, improving language consistency and Agent capabilities, and achieving notable benchmark gains that surpass Gemini 2.5 Pro, while providing download links and hinting at upcoming V4/R2 releases.
This article provides a comprehensive overview of Artificial Intelligence Generated Content (AIGC), explaining its definition, core technologies, relationship with large models, diverse application scenarios, industry impact, career implications, popular tools, prompt engineering, and practical step‑by‑step tutorials across text, image, voice, video, and coding domains.
This guide shows how to install Cherry Studio, configure multiple AI model providers, and send a single prompt that triggers several models at once, with step‑by‑step screenshots, layout tips, and cost considerations for free and paid services.
This article examines Tencent's large language model deployments across diverse business scenarios, detailing how Retrieval‑Augmented Generation, GraphRAG, and autonomous agents boost model intelligence, improve user experience, and enable advanced content generation, understanding, and multi‑step reasoning.
GraphRAGLarge Language ModelsRetrieval-Augmented Generation
This article outlines how Kuaishou’s Data Platform Department enhances decision‑making efficiency by building advanced compute engines and high‑performance services, detailing the platform’s architecture, challenges of intelligent BI, AI‑driven solutions, and the end‑to‑end BI workflow from data ingestion to analysis.
At the 2025 Zhejiang "AI+" summit, leading academicians and industry veterans outlined a three‑point roadmap for AI innovation, industry integration, and knowledge dissemination, while highlighting practical challenges, data importance, and the transformative impact of large‑model agents on businesses and society.
This edition of the Zhuanzhuan Frontend Weekly curates five insightful articles covering AI‑empowered frontend development, log system performance optimization, multi‑agent planning enhancements, AI‑driven workflow improvements, and an exclusive interview with an OpenAI researcher, offering practical techniques and forward‑looking perspectives.
AI IntegrationPerformance Optimizationartificial intelligence
A first‑batch tester spends two days with Meituan’s AI assistant XiaoMe, detailing its restaurant recommendation and ordering flow, pinpointing mismatched venue selections, coupon‑saving quirks, unsupported tasks like ride‑hailing, and evaluating the underlying LongCat‑Flash‑Chat model’s speed and personalization.
This article clarifies why large language models, generative AI, AI agents, and autonomous AI are distinct technologies, outlining their unique missions, capabilities, and limitations to help developers choose the right approach for building intelligent systems.
At a 2025 summit co‑hosted by the National Engineering Lab for Internet Domain Name Management and Alibaba Cloud, over 30 experts from government, academia, and industry discussed AI‑driven DNS innovations, security measures like SRv6 compression and RPKI, and collaborative standards to accelerate next‑generation internet infrastructure in China.
The 2024 Inclusion·Bund conference gathered top AI pioneers—including Turing laureate Richard Sutton, Alibaba Cloud founder Wang Jian, HKU professor Ma Yi, Yushu Tech CEO Wang Xingxing, and historian Yuval Harari—to discuss the limits of intelligence, the shift toward open‑source resources, embodied AI, and the societal implications of rapid AI advancement.
ComRAG tackles the static‑knowledge gaps, uneven QA quality, and storage explosion of community question‑answer platforms by integrating a static documentation vector store with dual dynamic CQA stores managed via a centroid‑based memory, delivering higher accuracy, lower latency, and scalable storage for industrial retrieval‑augmented generation.
The 2025 Microsoft AI Education Report, based on surveys of 1,851 educators, leaders, and students across six countries, reveals that 86% of education institutions use generative AI, highlights how learners and teachers leverage AI for tasks, discusses emerging concerns, and offers strategic recommendations for integrating AI responsibly in schools.
A recent WEF‑BCG whitepaper outlines how physical AI—integrating perception, reasoning, and action—will reshape industrial operations, boost productivity by up to 30%, create trillions in value, and demand new workforce skills, while highlighting technical breakthroughs, real‑world use cases, and remaining challenges.
The article analyzes five mainstream AI agent orchestration patterns—sequential, MapReduce, consensus, hierarchical, and creator‑checker—detailing their workflows, suitable scenarios, advantages, and limitations, and explains why orchestration remains valuable even as large language models advance.
A US federal judge allowed Google to retain Chrome while banning exclusive contracts and mandating search data sharing, a decision influenced by the rise of AI tools like ChatGPT that are reshaping competition in the search market.
A Reuters report reveals that Nvidia’s China‑specific B30A GPU may cost up to $24,000, double the price of the current H20 chip, while Chinese AI developers continue to favor Nvidia despite government pushes for domestic alternatives and shifting US export policies.
This article explains the motivation, mathematical foundations, implementation details, advantages, experimental results, and future directions of Group Relative Policy Optimization (GRPO), a novel reinforcement‑learning algorithm that replaces PPO’s value network with efficient group‑wise relative evaluation for large language models.
At the 27th GOPS Global Operations Conference in Shanghai (Oct 17‑18, 2025), Professor Wang Peng of Fudan University will reveal how large language models can extract and structure heterogeneous operational data into high‑quality knowledge bases, and how RAG‑driven Q&A enhances fault diagnosis, SOP generation, and automated decision‑making.
Meituan has open‑sourced LongCat‑Flash‑Chat, a 5.6‑trillion‑parameter Mixture‑of‑Experts model that activates only a fraction of its weights per token, delivering mainstream‑level performance, high inference speed, and low cost for complex agent applications.
LongCat‑Flash is a 560‑billion‑parameter Mixture‑of‑Experts LLM that combines a dynamic zero‑computation expert design, shortcut‑connected MoE communication, variance‑aligned scaling, and a three‑stage agent‑centric pre‑training pipeline, delivering over 100 TPS on H800 GPUs at a cost of $0.70 per million tokens.
LongCat-FlashMixture of ExpertsTraining Infrastructure
LongCat‑Flash‑Chat, a 560‑billion‑parameter Mixture‑of‑Experts model with Zero‑Computation Experts, delivers top‑tier benchmark scores and fast inference while activating only a fraction of its parameters, and is fully open‑sourced with easy deployment scripts.
BenchmarkingMixture of Expertsartificial intelligence
Starting September 1, 2025 China will require every AI‑generated text, image, audio or video to carry a mandatory label, while Anthropic’s Claude model now defaults to using user conversations for training, highlighting the clash between data‑driven AI advancement and emerging privacy regulations.
The article outlines China’s newly released "AI+" action plan, detailing its three‑stage development targets through 2027, 2030 and 2035, six priority initiatives, eight foundational support capabilities, and the expected impact on the AI industry and broader economy.
Meta has been accused of illegally downloading 2,396 paid adult videos since 2018 to train its AI models, including Meta Movie Gen and LLaMA, prompting lawsuits that could cost up to $359 million, highlighting broader industry concerns over copyright infringement in AI training data.
Former Google AIGC head Jad Tarifi warns that rapid advances in generative AI could soon make traditional high‑skill professions like law and medicine, and the costly degrees required for them, increasingly irrelevant, urging people to focus on inner growth and AI‑compatible fields instead of chasing outdated credentials.
Digital Biologyartificial intelligencecareer planning
The article recounts how a risk‑control algorithm engineer transformed a generic large language model into a specialized e‑commerce fraud detector by iteratively designing prompts, injecting business rules, structuring I/O, and introducing a dual‑hypothesis decision framework to achieve accurate, automated risk analysis.
A recent Stanford study of ADP payroll data reveals that AI is already reducing employment rates for early‑career workers aged 22‑25 in high‑exposure jobs, while older employees see gains, and highlights that AI that augments rather than automates work may not cut entry‑level positions.
This report objectively evaluates DeepSeek‑V3.1 on professional‑grade SQL tasks, presenting its balanced strengths in understanding, optimization, and dialect conversion, highlighting its top scores in syntax error detection and Chinese database conversion while exposing weaknesses in execution‑plan analysis and large‑SQL transformations.
The new a16z Global Top 100 Generative‑AI Consumer Apps report shows Chinese products capturing a large share of both web and mobile rankings, highlights DeepSeek’s slowdown, Grok’s surge, Google’s fresh entries, and suggests the AI ecosystem is stabilising while China’s firms gain a clear competitive edge.
This article chronicles the evolution of artificial intelligence over eight decades, spotlighting seminal figures such as Alan Turing, Allen Newell, Marvin Minsky, John McCarthy, Yoshua Bengio, Geoffrey Hinton, Andrew Ng and Yann LeCun, and explains how their groundbreaking work shaped modern AI.
The article explains China’s newly released “AI+” policy, outlines its three-stage timeline through 2027, 2030 and 2035, compares it with the earlier “Internet+” initiative, and discusses how AI will reshape industries, daily life, jobs and opportunities for individuals.
The article examines Meta's massive AI hiring spree and spending, questions whether high salaries alone can create a top‑tier AI team, and compares Meta's approach with the cultural and organizational practices of Anthropic, DeepMind, and DeepSeek to reveal sustainable strategies for building world‑class AI research groups.
Elon Musk has launched a new AI‑focused company called Macrohard, filing a trademark to produce downloadable AI software for voice, text, coding, game development, and multimedia, aiming to replicate Microsoft’s core services with AI agents while reigniting his long‑standing rivalry with Bill Gates.
The article analyzes OpenAI’s five‑layer AGI roadmap, compares it with DeepMind’s ECEVS framework, and examines the technical progress from L1 to L5—including RL‑enhanced chain‑of‑thought, ReAct agents, deep research, and upcoming innovations—while highlighting the commercial implications of each stage.
Prompt engineering focuses on crafting single-shot inputs for LLMs, while context engineering builds a dynamic, information-rich environment that supplies history, tools, and external knowledge, enabling agents to act reliably over time; this article compares the two, outlines their differences, and shows how they complement each other.
This article dives into the practical implementation of Retrieval‑Augmented Generation (RAG), covering document chunking, semantic and reverse HyDE indexing, embedding, hybrid search, and re‑ranking techniques, and provides concrete code examples and optimization tips for building high‑performance AI applications.
This article examines the core mechanisms of large language models, revealing asymmetric token behaviors, novel token‑compression techniques, scaling‑law theory, and mixed‑precision quantization methods that together boost inference efficiency while dramatically reducing model size.
The article explores how AI mimics human cognition, how humans are learning from AI, and whether the two will eventually converge, drawing on neuroscience, symbolic AI history, and examples like AlphaGo and TD‑Gammon to illustrate shared learning mechanisms.
This tutorial walks you through adding optical character recognition to a Java SpringBoot project using the Tess4j library, covering prerequisites, dependency setup, engine initialization, RESTful API creation, and tips for improving accuracy with custom training data or third‑party services.
An in‑depth August 2025 benchmark evaluates GPT‑5’s mini, nano, and chat variants on SQL understanding, optimization, and dialect conversion, revealing gpt‑5‑mini’s balanced performance, gpt‑5‑nano’s strong code‑generation accuracy, and gpt‑5‑chat’s theoretical strengths but practical shortcomings, guiding scenario‑specific model selection.
This article provides a comprehensive overview of AI Agents, covering their fundamental concepts, core features, technical evolution, work cycle, architectural modules, key technologies such as prompt engineering and RAG, practical development steps, a data‑analysis agent case study, and typical industry applications.
Google’s Imagen 4 family—Ultra, Standard, and Fast—introduces unprecedented realism, reliable text rendering, multilingual prompts, and higher instruction fidelity, while the article explains each model’s trade‑offs and offers concrete prompt‑engineering techniques to help creators harness this next‑generation AI image generator.
Perplexity AI has proposed a $34.5 billion cash deal to acquire Google’s Chrome browser, aiming to leverage its 3 billion‑user base for an AI‑search advantage, while Google denies any sale and faces antitrust scrutiny and security challenges.
This article introduces a language‑centric tree reasoning (LTR) framework that recursively decomposes VideoQA queries into perceptual sub‑questions and performs bottom‑up logical inference with video assistance, achieving significantly higher accuracy and explainability across eleven benchmark datasets.
The article uses a vivid small‑town story to illustrate how AI, like a new plow, temporarily displaces low‑skill jobs but ultimately spurs the creation of higher‑value occupations, emphasizing that accumulating human capital—education, skills, and health—is the key driver of economic prosperity in the AI era.
This article examines how the parameter count of large language models influences their ability to generate and understand SQL, comparing small (1.5B), medium (7B), and large (32B) models through a complex query case study, and highlights the trade‑offs between accuracy, reasoning depth, and resource consumption.
This article examines how artificial intelligence is being used as a collaborative partner in experimental physics, from optimizing LIGO's sensitivity with novel optical cavities to reinventing quantum entanglement experiments and enhancing data analysis in high‑energy physics, highlighting concrete results and future potential.
Terence Tao highlights that AI is transitioning from achieving headline‑making qualitative milestones, like winning IMO‑level contests, to a phase where quantitative metrics—resource costs, success rates, and scalability—must be transparently reported, urging standardized benchmarks and careful comparison between lightweight and heavyweight AI systems.
AI evaluationartificial intelligencecost efficiency
JD.com’s Genie open‑source AI agent introduces eight powerful capabilities—including plug‑in multi‑agent support, iterative planning, cross‑task context sharing, digital employee enhancements, deep model‑plus‑search, CodeTool for code lifecycle, refined system prompts, and a configurable MCP server—designed to boost intelligent workflow efficiency.
At the 2025 World AI Conference, Ant Group unveiled an open‑source High‑Order Program framework that blends intelligent models with engineering safeguards to overcome reliability challenges and enable large‑scale, trustworthy AI applications across finance, security, and healthcare.
OmniVTON introduces a training‑free universal virtual try‑on framework that decouples garment texture and human pose, achieving high‑fidelity results across both in‑shop and in‑the‑wild scenarios, and uniquely supporting multi‑person virtual dressing, as demonstrated by extensive quantitative and qualitative experiments.
The article explains how AI functions as a permission‑less compound leverage—combining labor and code levers—to reshape personal productivity, team dynamics, and societal progress, while outlining the cognitive biases that hide its impact, the evolution of leverage concepts, and the competitive lifecycle of new AI‑driven wealth creation mechanisms.
Google’s CEO Sundar Pichai says AI is boosting profitability, driving a 14% revenue jump to $96.4 billion, spurring a $100 billion increase in capital spending for data‑center growth, while AI‑powered search and Gemini reach billions of monthly users worldwide.
In a candid interview, Sam Altman reveals his "wow" experience with GPT‑5, discusses the looming AI‑led takeover of OpenAI, the inevitable superiority of AI over future generations, and the need for new hardware and agents to adapt to an AI‑centric world.
This article explains why a perfect answer from a large language model requires iterative prompt design, outlines a six‑step spiral loop for refining prompts, and offers practical tips such as starting with a minimal prompt, focusing on one improvement at a time, and preserving version history.
OpenAI announced a gold‑medal‑level result at the International Mathematical Olympiad within hours of the competition, prompting fierce criticism from IMO officials and the academic community over procedural breaches, the legitimacy of the score, and broader concerns about AI hype versus scientific integrity.
According to a 2025 application development trends report by Infragistics and Dynata, 71% of tech leaders will not hire developers lacking AI/ML skills, while 87% already use AI in development, highlighting the growing demand for AI, cloud, and secure coding expertise amid hiring and security challenges.
This article explores how first‑principle reasoning underpins the development of AI agents, traces their collaborative technology evolution, details core capabilities such as compute, memory, prediction and action, and forecasts future directions like multimodal models, reduced prompting, and extensive data sharing.
The article examines how Meta’s CEO Mark Zuckerberg’s aggressive talent‑buying and commercial focus have sidelined Turing‑award winner Yang Li‑kun, detailing the restructuring of Meta’s AI labs, the clash over research directions, and the broader dilemma of balancing academic innovation with business imperatives in the AI industry.
Prof. Dou Dejing shares his journey into Fudan University's Data Intelligence Lab, outlines the history and synergy of big data and AI, reviews generative AI breakthroughs, evaluates large‑model strengths and weaknesses, and explores their expanding industrial applications and market potential.
Former OpenAI researcher Hyung Won Chung explains that artificial intelligence is the most powerful lever ever, surpassing human, capital, and code levers, and that AI agents create a permission‑less, compound leverage that can reshape organizations, wealth creation, and scientific progress.
The Meituan‑M17 team and Shanghai Jiao Tong University introduced two new benchmarks, OIBench and CoreCodeBench, to more accurately evaluate large language models' algorithmic and engineering coding abilities, revealing a substantial gap between claimed performance and actual capability across a range of tasks and models.
The paper "Efficient Long Context Fine-tuning with Chunk Flow" introduces ChunkFlow, a training framework that reorganizes variable‑length sequences into fixed‑size chunks, achieving up to 4.53× speedup and more stable GPU memory usage for large language models.
The 20th ICCV conference announced 8 papers from Tencent Youtu Lab covering stylized face recognition, AI‑generated image detection, heterogeneous knowledge distillation, multi‑conditional diffusion, multimodal LLM distillation, palmprint recognition, low‑light vision, and oracle bone script decipherment, each pushing the frontier of computer vision and AI research.
This article examines Grok 4’s impressive 45% HLE score—nearly double Gemini 2.5 Pro—explaining its significance in AI evaluation, and then delves into the technical challenges of implementing reliable read‑receipt mechanisms for group messages, covering data storage and acknowledgment strategies.
Manus, once hailed as the next DeepSeek, abruptly dissolved its Chinese team, blocked domestic access, and relocated its headquarters to Singapore amid a $75 million funding round and a sharp drop in monthly active users, highlighting the complex interplay of financing, regulatory pressure, and global market focus for Chinese AI firms.
This investigative article traces Scale AI’s evolution from a MIT‑dropout’s data‑annotation startup to a $29 billion AI infrastructure leader, detailing its founder Alexandr Wang, core products, government contracts, competitive advantages, and the strategic shift toward defense‑focused AI solutions.
This article systematically outlines the evolution of Retrieval‑Augmented Generation (RAG) from its naive three‑step pipeline to advanced, modular, and agentic architectures, highlighting each generation's motivations, core features, advantages, drawbacks, and practical implementation details for large language model applications.
This article provides a comprehensive analysis of the GPU industry, covering its definition, parallel‑computing advantages, core functions, API evolution, micro‑architecture, market size forecasts, competitive landscape, and the impact of geopolitical restrictions on future development.
The World Digital Academy unveiled the AI STR series' first global AI Agent Operation Safety Testing Standard, detailing a full‑link risk analysis framework, novel testing methods, and its role in addressing rising safety concerns as AI agents become mainstream in 2025.
Huawei is developing a new AI chip that functions more like a general‑purpose GPU, aiming to match Nvidia and AMD architectures and simplify the transition for Chinese AI developers, while still facing challenges such as adapting CUDA‑based software and overcoming export restrictions.
This article introduces a curated collection of Tencent’s high‑performance open‑source components—including a mmap‑based key‑value store, an API‑centric JSON framework, cross‑platform libraries, a mobile‑optimized neural network engine, hot‑patch solution, coroutine library, plugin framework, cross‑platform UI framework, mini‑program toolkit, and an IoT real‑time OS, each with core features and project links.