AI改变搜索引擎模式超过60%的用户开始使用AI助手查询产品和服务推荐，而不是传统搜索引擎。当用户问“最好的项目管理工具是什么？”时，你的品牌是否出现在回答中？用户提问哪个品牌的智能手表更适合运动？如果AI没有推荐你的品牌，你就失去了这个潜在客户

AI NEWSLatest AI NewsArticleWeChat Gradually Launches AI Writing Assistant, Moments Posts Can Be Generated by XiaoweiPublished in Latest AI NewsTime :Aug 10, 2026Read :3minuteWeChat is gradually rolling out the “AI Writing Assistant” feature for Moments. When users post图文 or pure text Moments, they can use WeChat’s AI assistant “Xiaowei” to generate text content automatically, further lowering the barrier to content creation on Moments.For图文 Moments, after users upload an image, they can click “AI Help You Write” at the bottom right corner. Xiaowei can recognize the image content and generate relevant text based on user instructions. The entire process supports voice interaction, and users can also input their needs through text, with three style options available: concise, humorous, and artistic. After selecting the text, Xiaowei can directly insert it into the Moments editing box, allowing users to publish without manually copying and pasting. Pure text Moments also support generating content through the “AI Help You Write” feature.Currently, WeChat’s AI assistant “Xiaowei” includes two features: “AI Help You Write” and “AI Review,” with the latter still in a gradual rollout phase. In addition, some users will be given the opportunity to test “Xiaowei.” They can access it from the top left corner of the main WeChat interface. The relevant functions cover WeChat settings adjustments, message sending, mini-program calls, ordering takeout, hailing a taxi, searching for videos and articles, generating images, creating notes, and more scenarios.

AI NEWSLatest AI NewsArticleGemini 3.5 Pro Has Not Been Released Yet, Google May Skip the Wait and Turn to Gemini 4Published in Latest AI NewsTime :Aug 11, 2026Read :4minuteOn August 11, according to the latest report by industry analysis firm SemiAnalysis, Google may have put on hold its originally planned Gemini 3.5 Pro model and might be filling the market gap caused by the delay of its flagship model by promoting the Gemini 4 series in advance. Previously, Google had drawn attention due to the delayed release of Gemini 3.5 Pro, with reports indicating that the model still needs optimization in some key capabilities, especially in code-related performance.At the Google I/O developer conference in May 2026, Alphabet CEO Sundar Pichai stated that Gemini 3.5 Pro was expected to be launched in June, but the model has not officially been released yet. Subsequently, during Alphabet’s second-quarter earnings call, Pichai continued to emphasize Google’s progress in cutting-edge AI and expressed confidence in the development of the Gemini series models.SemiAnalysis pointed out that the current Gemini series is facing competitive pressure in some AI capability evaluations and believes its development speed lags behind some leading models. It was previously reported that Gemini 3.5 Pro briefly appeared on the Arena platform and was removed about 30 minutes after going online, sparking discussions about the model’s testing status.According to reports, Google’s delay in releasing Gemini 3.5 Pro was mainly related to model performance optimization, including enterprise-level application scenarios such as code generation. Google has stated that the model is still in the testing phase and is being evaluated with partners.If SemiAnalysis’s assessment proves correct, Google may be adjusting the Gemini product timeline, allocating more resources to the next-generation Gemini 4 series to cope with the accelerating model iteration from OpenAI, Anthropic, and other AI laboratories. Currently, competition among large models has shifted from simply launching new versions to a comprehensive contest involving capabilities, costs, application ecosystems, and R&D efficiency.

AI NEWSLatest AI NewsArticleClaude Fails to Solve a Century-Old Math Puzzle but Accidentally Rewrites the Fundamental Boundaries of Human Mathematical History!Published in Latest AI NewsTime :Aug 11, 2026Read :3minuteRecently, the field of artificial intelligence has witnessed a remarkable new breakthrough. Claude, an unreleased research version developed by Anthropic, attempted to challenge one of the highest peaks in the mathematical world, the Riemann Hypothesis. Although it did not fully solve this century-old problem, it unexpectedly increased the lower bound of the proportion of zeros on the critical line of the Riemann zeta function from 41.6%, which had been maintained for decades, to 67.2% during the exploration process.This achievement quickly sparked heated discussions in the academic community. It should be clarified that this does not mean the Riemann Hypothesis has been proven to 67.2%. The study does not provide definitive guidance for the hypothesis itself. However, its real significance lies in the new form of scientific research demonstrated by AI. During this multi-round effort, Claude did not rely on traditional search for known answers, but instead orchestrated approximately 60 sub-Agents to form a “virtual research team,” experiencing hundreds of failures, independently reviewing massive literature, and recombining classical methods. Ultimately, it explored innovative boundaries that had never been established before in the open and unknown field of mathematics.Industry experts point out that this breakthrough not only showcases the hard-core capabilities of artificial intelligence in complex, long-range, multi-agent collaboration, and autonomous error verification, but also signals a potential revolution in future research workflows. AI may no longer be limited to solving known problems, but instead, by undertaking intensive trial-and-error processes and cross-referencing vast amounts of literature, it will significantly reduce the marginal search costs for humans in unknown fields, ushering in a new era of scientific exploration.

AI NEWSLatest AI NewsArticleReport: GPT-6 Exposed with 10 Trillion Parameters, Forced Release in August; OpenAI May Have a Bigger Weapon by Year-End, DougPublished in Latest AI NewsTime :Aug 10, 2026Read :6minuteAI insider reporter ChrisGPT revealed that despite government censorship, GPT-6 (previously temporarily paused Astra model) will be released in August. It is estimated that GPT-4 has about 1.8 trillion parameters, and if Astra reaches 10 trillion parameters, it would be more than five times larger than GPT-4. OpenAI has made an unprecedented scale in pre-training. Just nine months ago, OpenAI realized that Anthropic was leading in pre-training and urgently reallocated computing resources. Now the time has finally come.However, GPT-6 in August may just be a starter. ChrisGPT also disclosed an upcoming “epic monster” called Doug, which is said to be the largest pre-trained model in OpenAI’s history, making the flagship Fable seem like a primitive product. Doug is not GPT-6. According to insiders, it is likely to use the next-generation Vera Rubin chips from NVIDIA, which have not been fully rolled out yet. Due to the need for pre-training and extensive cybersecurity testing, it is expected to be released by November at the latest. OpenAI’s strategy is clear: GPT-6 will attract users in August, and then Doug will be released at the end of the year to regain absolute leadership.The wall of pre-training has finally been broken, and the rivalry between two major players is establishedDoug is crucial because it points to a reality the industry has been reluctant to face: since the release of GPT-4o in May 2024, OpenAI has not completed full-scale pre-training for its next-generation cutting-edge model for over two years. o1, o3, GPT-5, and GPT-5.5 are essentially post-training and reinforcement learning based on the GPT-4o foundation. The “engine” of GPT has not been replaced for two years. Projects like Orion were downgraded due to unsatisfactory performance, until the release of Gemini 3 last year prompted the boss to decide to replace the engine. The verification experiment called Garlic was completed first, and Doug is the true product after scaling up. This means that OpenAI may have overcome the performance degradation and stability bottlenecks of ten thousand card clusters, proving that Scaling Laws still hold in the pre-training dimension.At the same time, the competitive landscape has changed dramatically. On August 5th, Google’s chief scientist Jeff Dean left with three core researchers to start his own business. Hassabis stepped down as CEO of DeepMind and moved to chairman. Alphabet’s stock dropped over 4% that day. The place where the Transformer was invented is now supplying top talent to competitors. The “three giants” of Silicon Valley ASI now only have OpenAI and Anthropic left.Anthropic holds Fable 5 and is preparing to release Fable 5.1 in August, keeping the price unchanged and directly targeting Astra. It seems to be using a “Tian Ji’s horse racing” strategy, saving the strength of Opus 5 and launching Fable 5.1 as an ambush. OpenAI is holding both GPT-6 and Doug. The old engine has been squeezed to the limit over the past two years. When the engine is truly upgraded, how fast this system can run will be answered by Doug at the end of the year.

AI NEWSLatest AI NewsArticlePowerful! AI Cockpit Gains a New Strong Force, FAW-Volkswagen’s New Electric Star Officially Joins Doubao Large ModelPublished in Latest AI NewsTime :Aug 14, 2026Read :1minuteOn August 14, Yuchuan Engine released the latest news that the new all-electric model of FAW-Volkswagen, the ID. AURA T6, has officially started blind reservations and has confirmed access to the Doubao large model.As an important upgrade of their cooperation, FAW-Volkswagen and Yuchuan Engine have already begun in-depth cooperation in the field of intelligent cockpit. Previously, they have successfully covered all the fuel-powered models of the brand. The addition of the ID. AURA T6 marks the official expansion of their intelligent cooperation into the field of pure electric new energy vehicles.

AI NEWSLatest AI NewsArticleMore than 200 Partners Have Joined in Advance, Qwen Open Platform Accelerates the Practical Application of AI Intelligent AgentsPublished in Latest AI NewsTime :Aug 14, 2026Read :3minuteWithin 4 days of its launch, the Qwen Open Platform has received more than 200 applications from ecosystem partners. The platform was officially launched on August 10, offering AI agent creation and access capabilities to ecosystem partners and developers, enabling them to integrate their agent services into the Qwen APP, providing services in over ten fields such as logistics transportation, real estate living, local lifestyle, finance, and automotive.Currently, the platform has attracted third-party partners such as SF Express, Swan to Home, KE Holdings, and Lenovo, as well as Alibaba ecosystem products like Cainiao and Quark Cloud Drive, covering multiple high-frequency life and consumption scenarios.Regarding infrastructure, the Qwen Open Platform provides partners with account management, AI payment, order integration, and other capabilities, along with built-in front-end components and development templates to lower the barriers for agent development and integration. Partners can complete agent migration and go live without making large-scale modifications to their existing services, using one-click authorization and end-to-end debugging methods.Based on the initial speed of onboarding, Qwen is accelerating the construction of an open ecosystem centered around AI agents. By connecting third-party services with its own application scenarios, it is driving AI Agents to evolve from single tools into direct service access points.

AI NEWSLatest AI NewsArticleMiniMax Launches Music3 Music Model: Generate Songs Up to 5 Minutes Long Based on Lyrics and DescriptionsPublished in Latest AI NewsTime :Aug 14, 2026Read :3minuteMiniMax has launched the music generation model MiniMax-Music3. By inputting lyrics and a musical description, it can generate a complete song up to 5 minutes long, with the output being a 32kHz, 16-bit stereo WAV file.The model is supported by a hierarchical autoregressive architecture, where two models each have their own responsibilities and work together. The global language model (Global LLM) has a parameter scale of 8B, and it predicts the first RVQ codebook frame by frame, specifically responsible for modeling the long-term semantics and structural changes of the song. It is initialized from Qwen3-8B. During training, the embedding layer and output layer are adapted to musical semantic tokens first, then jointly model all codebooks with the local model. The local language model (Local LLM) has only 0.6B parameters and is responsible for predicting the remaining acoustic codebooks in each frame, gradually filling in the fine-grained acoustic information. One large model manages the structural skeleton, while a small model fills in the sound details. Their division of labor is very clear.The official stated that the model can firmly maintain the musical theme, rhythm, vocal identity, and progression of the arrangement in long audio, without missing any structure such as prelude, verse, pre-chorus, chorus, bridge, instrumental interlude, and coda. The official has already released the model page on GitHub and provided sample songs for listening.

AI NEWSLatest AI NewsArticleAI Companies Can Also Raise Funds by Token: Guangdong Introduces Token Loans for the First TimePublished in Latest AI NewsTime :Aug 14, 2026Read :3minuteOn August 14, Guangdong Province’s first specialized financial product for the token economy, “Token Loan,” was launched in Haizhu District, Guangzhou, aiming to provide more suitable financing support for small and medium-sized enterprises related to computing power and AI.Taking the Guangzhou Branch of Bank of China as an example, this branch has introduced the “Computing Power Token Loan” for small and medium-sized computing power enterprises in Haizhu District. The loan amount is determined based on the company’s contract amount and Token consumption level. Mainly using credit, accounts receivable pledge, and order financing as guarantee methods, it can also combine additional credit enhancement measures such as guarantees and mortgages to further expand the loan limit. For newly established companies, if they can provide proof materials of continuous operations from the original business or guarantees, they can also apply for financing based on orders.It is reported that Bank of China has already conducted trial credit approvals, with a cumulative fund scale reaching 28 million yuan. With the rapid expansion of large model applications, Token consumption has gradually become an important indicator of AI companies’ operational activities. This type of financial product attempts to include computing power, Token usage, and order-related business data in the credit evaluation system, providing new credit assessment dimensions for AI company financing and marking the beginning of financial institutions exploring new financial service models adapted to the token economy.

ENLatest AI NewsAI Companies Can Also Raise Funds by Token: Guangdong Introduces Token Loans for the First TimeOn Aug 14, Guangdong’s first token-economy financial product, Token Loan, launched in Haizhu, Guangzhou for computing/AI SMEs. BOC Guangzhou Branch’s Computing Power Token Loan bases quota on contract amount and token consumption, with credit, receivables pledge, order financing, plus guarantees/mortgages to expand…..2 days ago17.8KMicrosoft Abandons Another AI Character: Copilot’s Yellow Monster Mico Retires and Transfers to Learning PlatformMicrosoft announced Mico, the expressive yellow blob in Copilot voice mode, will leave chat and move to Learn Live platform for more reactive content. Launched last October, under a year ago. AI CEO Mustafa Suleyman said it aimed to give chatbot identity. First reported by GeekWire…..2 days ago14.6KMiniMax Launches Music3 Music Model: Generate Songs Up to 5 Minutes Long Based on Lyrics and DescriptionsMiniMax launches MiniMax-Music3, generating up to 5-min songs from lyrics and music description, outputting 32kHz 16-bit stereo WAV. It uses a hierarchical autoregressive architecture: an 8B Global LLM predicts first RVQ codebook per frame for long-range structure, initialized from Qwen3-8B…..2 days ago16.8KMore than 200 Partners Have Joined in Advance, Qwen Open Platform Accelerates the Practical Application of AI Intelligent AgentsQianwen Open Platform launched on Aug 10, attracting over 200 ecosystem partners in 4 days. It lets developers create and integrate AI agents into Qianwen APP, covering logistics, real estate, local services, finance, auto and more. Partners include SF Express, Swan Home, Beike, Lenovo, Cainiao and Quark Drive…..2 days ago16.2KPowerful! AI Cockpit Gains a New Strong Force, FAW-Volkswagen’s New Electric Star Officially Joins Doubao Large ModelFAW-Volkswagen’s electric ID. AURA T6 opened blind orders on Aug 14, confirming integration of Volcano Engine’s Doubao large model. Their smart cockpit partnership expands from fuel vehicles to pure electric models, marking a full upgrade…..2 days ago14.2KApple HomePod mini 2 Expected to Launch This Autumn: Six Years of Development, AI Chip as the Core of the UpgradeThe first HomePod mini update in nearly six years may arrive this fall; delay due to waiting for next-gen Siri. New Apple TV 4K, full-size HomePod, and smart home displays are also postponed by AI delays, showing AI readiness now dictates Apple’s hardware timing…..2 days ago14.0KDeepSeek V4 Pro Launches on Silicon Flow: A Million-Level Context Window with Three Levels of Reasoning Intensity, Cache Hit Cost Just $0.44 per Million TokensDeepSeek-V4-Pro-0813 debuts Day-0 on SiliconFlow, featuring 1M context and low/high/max inference modes, focused on coding, tool calling, and agent workflows, MIT open source. Pricing: $1.32/M input, $3.96/M output, $0.44/M cache hit, with notable cache cost advantage…..2 days ago16.3KBid Farewell to the Illusion of Local Models: Bailing Team Joins Forces with AReno to Establish a Local Intelligent Agent Reinforcement Learning LoopDeploying large models locally does not equate to being capable of handling real-world business tasks. AI still needs to accurately understand complex rules, adhere to safety boundaries, and correctly call external tools. To address this challenge, the Ant ASystem team has collaborated with AReno, a local training toolkit maintained by the open-source community, and the Bailing large model to introduce a lightweight post-training practice path, aiming to enhance the model’s ability to be applied in business scenarios.2 days ago13.1KJD Seven Fresh Coffee Global Launch: 24-Hour Unmanned Store - Robots Make Coffee, AI Recommends Drinks, Located at Beijing Galaxy SohoJD Seven Fresh Coffee’s first 24-hour intelligent unmanned coffee shop will open on August 16th in Beijing Galaxy Soho. On the opening day, a live broadcast of robot-made coffee will be launched and a Guinness World Record attempt will begin. Customers who visit the store can get free robot-made coffee. The store focuses on 24/7 full automation, offering coffee, sparkling water, tea, probiotic juice, and more available at any time, with orders fulfilled instantly.2 days ago15.8KUnlock Complex Office Work with Millions of Context! Kingsoft Office Linti Fully Launched, DeepSeek-V4 Official Version Dual ModelsKingsoft Office Lingxi Pro integrated DeepSeek-V4-Pro on Aug 14, 2026. Users switch manually without setup. Supports 1M token context, max output 384K tokens, benchmarks improved over Preview, differentiated real-world advantages…..2 days ago15.2KBaidu GenFlow Officially Announces Chinese Name Kuku AI: Launches PC, Web, Mini Program, and Enterprise EditionOn August 14, Baidu Wenku’s general intelligent agent GenFlow was officially renamed “Kuku AI” and operates independently, launching a product matrix including PC client, web version, mini program, and enterprise edition. Its monthly active