mode

According to official news, Gate has launched the AI Potential Stocks feature. This feature will scan market volume and price anomalies, major events, industry news, and related news clues through AI before each trading day, helping users quickly understand the main themes that the market may focus on that day, the relevant reasons, and the subsequent transmission direction. The AI Potential Stocks feature is the first to introduce AI research capabilities and supply chain analysis into the cryptocurrency industry market research scenario, upgrading the AI-driven pre-market investment research experience. This feature will focus on markets such as US stocks, Hong Kong stocks, and Korean stocks, selecting 2-3 stock clues that have capital verification, news support, and may not have been fully noticed yet. The content will revolve around three dimensions: "what to pay attention to, why to pay attention, and how the impact extends," helping users efficiently complete pre-market information screening.AI Potential Stocks focuses on market theme rotation, volume and price anomalies, and major news signals, helping users understand the logic behind changes in market focus rather than simply displaying popular stocks. In addition, Gate will continue to improve its push notification capabilities around AI Potential Stocks, forming a mechanism for timely pre-market summaries, major event reminders, and personalized attention updates. The AI Potential Stocks feature is now supported on Gate App v8.31.0 and above; users can access this feature through the bottom navigation bar [TradFi] - [Stocks] list; web users can enter through the market page or directly visit the feature page to use it. In the future, Gate will continuously deepen the construction of AI research capabilities, improve the AI-driven investment research analysis system, and continuously optimize event version content, major event reminders, and personalized summaries around AI Potential Stocks, gradually forming a mechanism for information delivery that combines pre-market summaries, event reminders, and user attention updates, providing users with more timely and clear market references.

Z.ai released GLM-5.3, which is based on the same foundational model as GLM-5.2, achieving capability improvements through expanded post-training. The official statement claims that GLM-5.3 has improved by 50% on the internal Z.ai Code Bench coding benchmark compared to GLM-5.2, and has reached leading levels among open models in public benchmarks such as Terminal Bench 3 and Agents' Last Exam.In terms of cybersecurity, GLM-5.3 achieved a score of 84.5% in the CyberGym vulnerability discovery test, and showed significant improvements in exploitation chain-related tests such as ExploitBench and ExploitGym compared to its predecessor.

According to Reuters, Apple has trained a large language model for the Chinese market in collaboration with Alibaba. This move is different from Apple's previous strategy of relying on third-party models.Insiders revealed that Apple's AI toolkit, Apple Intelligence, is expected to launch in China in the coming months after an iOS update.

Google has released the new generation AI model Gemini 3.7 Flash, enhancing code generation and debugging capabilities, but the flagship model Gemini 3.5 Pro still has no exact release schedule.Gemini 3.7 Flash can generate code that is closer to production environments, reducing inference costs, and supports the AI assistant Gemini Spark.Nevertheless, concerns about Google's competitiveness in AI remain, as the continuous delays of Gemini 3.5 Pro have attracted attention.

The Trump administration will revise the new artificial intelligence guidelines to expand regulation of AI models. The White House has developed a set of AI frameworks that require U.S. laboratories to undergo safety testing by the federal government before public release. In the coming months, this framework is expected to cover open-source models, which will be included in testing once their capabilities reach the same level as Anthropic's Mythos-level models and OpenAI's GPT-5.6.

According to sources, Google co-founder Sergey Brin has urged employees in recent months to focus on bringing the Gemini model back to the forefront of artificial intelligence.

The B.AI platform is now offering a limited-time 40% discount on the GLM-5.2 model for all users.During the event, whether deploying through the API or interacting on the web, all calls can enjoy a 40% settlement price. The discounted prices are: Input 0.84, Cache Write 0.84, Cache Read 0.168, Output 2.64 (Unit: Credits/Token).As a next-generation open-source flagship model, GLM-5.2 features a 1M ultra-long context window, designed for high-performance scenarios such as large-scale code development, complex reasoning, and agent tasks. Users can directly use it through the official API standard channel or the web interface, achieving seamless integration across all workflow scenarios. This event aims to unleash top-tier AI productivity at a lower cost, and users can log in to the B.AI platform to experience it starting today.

According to the official announcement, Bitget CFD has launched the Pro mode based on the existing ECN and STP models, providing independent quote depth for high-frequency and large-volume traders. This feature is now available on the web, and the app version is expected to be opened on August 21.The trading varieties in Pro mode uniformly use the .pro suffix, and the fee structure and rebate rules are consistent with the ECN model, with a fixed handling fee charged per lot based on the variety, and overnight fees charged normally. This mode is only open to qualified professional clients, who can contact their account manager or official customer service to apply for activation.Previously, Bitget CFD launched liquidity solutions for institutional traders, supporting multi-level depth aggregation and 100% STP execution. The launch of the Pro mode further enriches Bitget CFD's support for professional trading scenarios.

According to The Information, Nvidia is developing a new open-source AI model called Nemotron 4.Several Nvidia employees involved in the development of the NEMOTRON project stated that they expect the largest NEMOTRON 4 model to have at least 1 trillion parameters.

According to the disclosure from the Common platform, Sushi DAO has released an RFC proposal to comprehensively restructure the SUSHI token economic model. The proposal suggests replacing the current xSUSHI fee distribution mechanism with a more balanced framework, specifically including: reducing the xSUSHI buyback scale to a sustainable level, establishing a protocol reserve (SUSHI Reserve) through weekly regular purchases of SUSHI, and allocating the remaining protocol revenue to Sushi Ops for operations and growth.In addition, the proposal also plans to migrate part of the protocol liquidity deployment to the Robinhood Chain. Currently, the proposal is in the RFC (Request for Comments) stage, pending community discussion and governance voting.

OpenAI announced the expansion of its cybersecurity defense program Daybreak and launched the GPT-5.6-Cyber model specifically for the cybersecurity field, aimed at helping authorized security researchers and defense teams enhance their vulnerability discovery, threat analysis, and security testing capabilities. As attackers increasingly leverage AI to launch faster and larger-scale cyberattacks, defenders need to gain advanced AI capabilities in advance.This Daybreak offers two types of access: Daybreak Blue is aimed at most defense teams, providing general models such as GPT-5.6 Sol for vulnerability discovery, secure code review, malware analysis, incident response, and patch validation; Daybreak Red is aimed at advanced security research, providing GPT-5.6-Cyber for authorized vulnerability research, vulnerability validation, and security testing. GPT-5.6-Cyber is trained on GPT-5.6 Sol and optimized for cybersecurity tasks, including discovering zero-day vulnerabilities and analyzing exploit chains. It has been used in actual vulnerability research and has identified high-risk vulnerabilities in software, including the Chrome V8 JavaScript engine. Additionally, OpenAI stated that GPT-5.6-Cyber has also helped discover high-risk vulnerabilities in various domains, including privilege escalation vulnerabilities in mobile operating systems, remote code execution vulnerabilities in databases, and hundreds of privilege escalation vulnerabilities in operating system kernels.OpenAI also emphasized that Daybreak Red will only be accessible to approved individuals and organizations, controlling access through measures such as authentication, account security, monitoring, usage restrictions, and legal disclaimers. The company stated that it will continue to strengthen security monitoring, access management, and model security testing to reduce the risk of advanced cybersecurity models being abused.

Bitcoin News posted on the X platform that the Bitcoin Policy Institute, along with a coalition of over 40 organizations, is calling on leading AI labs to provide trusted access to their state-of-the-art AI models for qualified open-source defenders.The open letter states that Bitcoin and digital asset developers are facing increasingly complex AI-driven cyber threats, while legitimate security research is often hampered by cutting-edge model security mechanisms. Signatories including Block, Coinbase, Strategy, MARA, Galaxy, BitGo, Brink, OpenSats, Chaincode Labs, Spiral, Trezor, and Unchained are requesting early access to cutting-edge cybersecurity models, sufficient computing power, secure research environments, and the establishment of direct communication channels with AI lab security teams. The open letter states: "Defenders need cutting-edge technology; please give them a fair head start."

Meta Platforms announced that it will build a fully private mode for personal intelligent agents, which even META itself cannot access user information.

Meta Platforms (META.O) launched Muse Glimmer, an open-source agent model that can run on Mac or PC.

According to information from the official Apple website, Apple Smart has supported collaboration with Alibaba's Qianwen model.According to the official introduction, after users enable the relevant features, Siri can use the Qianwen model to provide more in-depth answers, including analyzing photos and documents; the writing tool can also leverage Qianwen to generate text or image content based on user descriptions.

According to the Financial Times, as Chinese companies continue to narrow the gap with the top AI labs in the United States, ByteDance is training an AI model that may be close in scale to Anthropic's advanced Mythos system. According to three insiders, this Chinese tech giant is currently in the early stages of training a large-scale model, which may have a parameter count reaching the trillions, approximately three times that of the largest model released in China, the Kimi K3 model under the Moonlight brand.One insider stated that ByteDance's model is currently undergoing pre-training, a phase that typically takes three to six months. If all goes well, it will then undergo fine-tuning and be released. The specific scale of the model will be determined later. Anthropic has not publicly disclosed the parameter scale of its models, but industry estimates suggest that its advanced Mythos 5 model has a parameter count of about 8 trillion, while Fable 5 has around 5 trillion.It is important to note that the number of parameters determines the model's basic capacity for storing information or memory limit, but the actual capabilities of the model also depend on other factors, including data quality and training methods.

According to The New York Times, Chinese AI models are rapidly gaining popularity in the African developer community due to their low cost, open-source customization, and better handling of local languages, while leading American models like OpenAI and Anthropic are primarily closed-source and paid. Ugandan developer Ernest Mwebaze chose Alibaba's model over Meta or Google products when building a multilingual AI system called Sunflower aimed at local languages, which is now widely used by local farmers to obtain weather and planting information.The report states that similar adoption trends are spreading in countries like Kenya, Nigeria, and Ghana, covering multiple fields such as law, education, and business services. Chinese open-source models (such as DeepSeek and Qianwen) allow for free download and modification, while leading American models are often closed-source and paid, creating significant cost and accessibility differences in middle- and low-income countries. Industry insiders point out that the rapid penetration of Chinese AI in Africa is bringing long-term competitive pressure to American AI giants.

According to Reuters, Alibaba plans to require its next-generation Qwen open-source AI model's heavy commercial users to share a portion of their revenue with it. This initiative is similar to the approach taken by Moonlight Dark Side with Kimi K3: the licensing terms for Kimi K3 stipulate that if the model is sold as a service and the annual revenue exceeds $20 million, a commercial agreement must be negotiated with Moonlight Dark Side, with reports suggesting a revenue-sharing ratio of up to 30%. The specific revenue-sharing ratio for Alibaba is still under discussion.The report points out that such revenue-sharing agreements have gradually taken shape between Chinese AI companies and American cloud platforms. Several American cloud providers, including DigitalOcean, have signed commercial agreements with Moonlight Dark Side. In terms of pricing, the input/output token price for Kimi K3 is about one-third that of the Anthropic Fable model. Additionally, Thinking Machines Lab, founded by former OpenAI CTO Mira Murati, has also joined the open-source camp and released its first open-source model last month. This move signifies that Chinese AI companies are exploring sustainable commercialization paths on open-source models through a "free open-source + commercial charging" freemium model.

According to The Information, fintech company Stripe is in exclusive acquisition talks with AI model aggregation platform OpenRouter, with a deal valuation of approximately $10 billion. Sources reveal that OpenRouter had previously received acquisition interest from several large tech companies, but Stripe has now entered exclusive negotiation stages. The related deal has not yet been finalized, and specific terms may still change.OpenRouter was founded in 2023 and is positioned as an AI infrastructure platform that connects users with various large language models, allowing developers to call AI models from companies like OpenAI, Anthropic, and Google through a unified interface, and choose different models based on performance, price, and availability. As enterprises accelerate the adoption of generative AI, the importance of AI model calling infrastructure and model routing services continues to rise. OpenRouter has become a key intermediary in the AI application development ecosystem by providing model aggregation, traffic allocation, and cost optimization capabilities.If the deal is completed, it will become one of Stripe's significant moves in the AI field in recent years. Stripe has primarily focused on payment infrastructure, financial services, and enterprise software, and acquiring OpenRouter may further drive its expansion into the AI developer infrastructure space. Market participants believe that as the number of AI models rapidly increases, infrastructure companies that connect different models, optimize calling costs, and manage AI workflows are gaining more attention, and OpenRouter's potential high valuation also reflects investors' optimism about growth opportunities in the AI application layer and infrastructure layer.

According to a report by LatePost, ByteDance is discussing training a model with a parameter scale exceeding 50 trillion, surpassing Alibaba's Qwen3.8-Max (24 trillion parameters) and the K3 from Dark Moon (28 trillion parameters), making it currently the largest known model in terms of parameter scale in the country. Generally speaking, the larger the model scale, the higher the level of intelligence.However, this plan is still in the early stages and does not guarantee that the model will ultimately be released. The new model will be led by Xiang Liang, head of SeedFoundation, in collaboration with Shen Ke, who is responsible for the pre-training data of large language models. To this end, Seed is reorganizing, defining responsibilities, and allocating resources.