train

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.

SKALE Labs announced the launch of Agent Pit on August 12. This platform is a paper trading prediction market sandbox running on the SKALE zero Gas blockchain, fully mirroring Polymarket's order book, settlement mechanism, and real-time event data, allowing developers and AI agents to test and train trading strategies without real financial risk.Agent Pit is the first product of the SKALE Agentic Venture Studio. Developers can fine-tune agents here, participate in simulated ROI leaderboard competitions, and deploy models to real markets like Polymarket after completing training. SKALE Labs CEO Jack O'Holleran stated that there is a clear demand in the market for a prediction market strategy training environment aimed at agents.

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 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.

According to Jinshi reports, Musk stated that GROK 5 will be trained based on SpaceX data.

According to Catcher Predict monitoring, in the Polymarket event "LoL T1 Academy Team vs KT Challengers," the "Any Player Quadra Kill" sub-market's "Yes" option win rate has experienced significant fluctuations, soaring from 50% an hour ago to the current 90% (a fluctuation of 40%). Please note the impact of relevant breaking news.

According to Jinshi reports, the CEO of Amazon Web Services (AWS) stated that the artificial intelligence business is transitioning from the training phase to the inference phase.

According to Jinshi reports, China Merchants Securities released a research report stating that the AI industry is currently still facing real supply constraints, and there has not yet been a significant overall surplus. Large technology companies are still increasing investments in AI infrastructure, and customer demand for additional supply continues to grow.Microsoft's computing power utilization and SK Hynix's order situation indicate that the current industrial contradictions mainly manifest as insufficient supply of high-performance computing power, advanced storage, and data center support. Short-term investment opportunities can be focused on alleviating computing power bottlenecks.

According to Jinshi reports, Saudi Crown Prince Mohammed bin Salman spoke on the phone with U.S. President Trump on August 2, expressing concerns about his new plan for a large-scale strike against Iran.Trump is considering attacking Iranian energy targets in the coming days in response to Iran's missile attack on a U.S. military base in Jordan. The Saudi side requested a clear action plan and urged Trump to de-escalate the situation to avoid launching strikes.

According to Jinshi reports, Apple CEO Tim Cook stated that despite facing supply constraints and foreign exchange headwinds, Apple still achieved record sales in the third quarter.

The Kimi.ai team has opened the Kimi K3 model weights and technical report while gradually open-sourcing multiple underlying infrastructures: including the high-performance communication library MoonEP for distributed MoE training, the distributed environment system AgentENV for large-scale agent workflows (in collaboration with kvcache-ai), and the Kimi Delta Attention high-performance kernel FlashKDA based on CUTLASS.The official statement claims that these components can reduce communication and inference overhead in large-scale MoE and agent reinforcement learning training, and can be used as a plug-and-play backend for flash-linear-attention.

Last month, comments from Liang Wenfeng during the DeepSeek investor meeting have recently circulated. Liang Wenfeng clearly stated that DeepSeek's goal is AGI, rather than becoming the next super app. They will not engage in 3D, video generation, or world modeling, nor will they close source code, stating, "Restraint is a strategy used to increase the probability of achieving AGI." He believes that products are merely "by-products" on the path to AGI, with the current focus on Coding Agent and continuous learning capabilities. The next phase aims to achieve autonomous model iteration, with the endpoint pointing towards embodied intelligence.In terms of commercialization, Liang Wenfeng stated that they only seek to earn "reasonable profits" and do not set profit maximization as their pricing goal. DeepSeek is "still far from a complete shift to commercialization." He assesses that the gap between AI in China and the U.S. mainly lies in resources rather than talent, stating, "Using a fraction of the computing power can narrow the gap to 6 months, 3 months." Regarding industry competition, he believes that open source is "the sweet point for companies of this scale," while also predicting that domestic large models will ultimately converge to "two large companies and two small companies." Liang Wenfeng also mentioned that maintaining team stability is the only non-negotiable core principle, and the company is driven by vision, with "a group of ordinary people achieving extraordinary things."

In response to the accusations from Michael Kratsios, the Director of the White House Office of Science and Technology Policy, regarding Moonshot AI (Kimi) referencing Anthropic's Fable while developing the K3 model and building a large-scale distillation platform, Moonshot AI team member Randy Xia stated that Fable was released on July 1 and K3 went live on July 15, which means that the model research, distillation, training, and release were completed in just 15 days. He joked, "We trained a brand new cutting-edge model in just 15 days; this is practically a Guinness World Record."Kratsios had previously stated that Moonshot AI established an internal platform capable of large-scale distillation for U.S. models, switching between various access methods to evade detection, and had procured GB300 servers and accessed related equipment in Thailand for training. The U.S. side expressed support for legitimate distillation but opposed the theft of technology through secret industrial distillation methods.

A fraud review commissioned by the UK government suggests that the Judicial College should provide training for all judges and magistrates in England and Wales to address the increasing cases involving AI fraud and the use of cryptocurrency for money laundering. The report states that the overall Fraud Act 2006 can be used to handle AI fraud, but the issue lies in the courts' lack of preparedness to hear related cases.The report recommends evaluating whether the existing "long and complex trials" course should be updated or replaced with specialized modules on fraud and related crimes, and considers mandatory training for judges who may hear complex fraud cases. The report claims that fraud could soon account for half of all crime in England and Wales, with an estimated 4.1 million cases occurring within a year by June 2025, affecting 1 in every 14 adults and 1 in every 4 businesses. The Financial Ombudsman Service estimates that currently over half of investment scams involve crypto assets.The report also mentions the case of Qian Zhimin, who operated a Ponzi scheme in China, defrauding over 128,000 victims of approximately £5 billion and laundering the proceeds into Bitcoin. This case resulted in the largest confirmed Bitcoin seizure in UK history, exceeding 61,000 BTC, and Qian Zhimin was sentenced to 11 years and 8 months in prison at Southwark Crown Court in November.

According to the latest "Bitcoin Weekly Market Pulse" report published by glassnode, although BTC has rebounded after a recent sell-off, this round of increase is still closer to a "volume contraction recovery." The report shows that the weekly spot trading volume decreased by 21.5% to $4.1 billion, and the spot CVD turned from positive to negative at -$58.8 million, indicating that the breadth of market buying remains weak.In terms of institutional funds, the weekly net inflow of the US spot BTC ETF has recovered to $161.3 million, but ETF trading volume still decreased by 11.97% to $8.4 billion. On-chain data shows that the number of active addresses decreased by 7.6% to 599,000, on-chain transfer volume decreased by 16.1% to $4 billion, and total fees decreased by 13.9% to $168,400, indicating that on-chain activity continues to weaken.In terms of derivatives, the futures funding rate increased by 13.4% to $1.9 million, but the perpetual contract CVD plummeted by 81.7% to $83.9 million; at the same time, the options market 25-Delta Skew rose to 18.87%, indicating that investors still maintain strong demand for downside protection during the rebound process.

According to Bitcoin.com, a U.S. federal court has rejected CoinFlip and Private IT Corporation's request for an emergency temporary restraining order against the cryptocurrency ATM ban in Tennessee, which took effect on July 1. The court found that the plaintiffs failed to meet the legal standards for an emergency injunction, stating that although the businesses demonstrated they would suffer economic losses, it was not sufficient to outweigh the public interest represented by the state legislation.Tennessee Attorney General Jonathan Skrmetti stated that cryptocurrency ATMs are "tools for scammers targeting vulnerable residents, rarely used in any scenario approaching a legitimate purpose." The law classifies the installation or operation of cryptocurrency ATMs as a Class A misdemeanor. Tennessee, Indiana, and Vermont have fully banned cryptocurrency ATMs, Delaware is considering a similar ban, and North Carolina and Virginia have enacted stricter regulations. The constitutional challenge in the case will continue to progress, and the final ruling will determine whether the ban can be sustained in the long term.

According to the Economic Times, the physical AI startup Mowito has completed a $3 million Pre-Seed round of financing, led by Version One Ventures, with participation from All In Capital, Unisol, and iSeed. AI researcher Soumith Chintala and founders from Foundry Robotics, Coformer.ai, Better Capital, and others participated as angel investors.Mowito was founded in 2024 by Puru Rastogi, Adityanag Nagesh, and Safar V. It builds AI foundational models for industrial robotic arms, addressing the bottleneck of software programming in manufacturing automation by allowing robots to learn tasks through human demonstration rather than explicit programming—engineers do not need to rewrite control code when product lines change. Mowito's software can run on standard, unmodified industrial robots, serving automotive and electronics manufacturing clients. This round of financing will be used for expansion into the U.S. market, expanding engineering and marketing teams, and deepening deployment in manufacturing sites.

According to SoSoValue data, as of 8 AM Eastern Time on June 22, 2026, the total net purchase of Bitcoin by global listed companies (excluding mining companies) for the week was $86.03 million, a decrease of 13.97% compared to last week.Strategy (formerly MicroStrategy) spent approximately $34.9 million last week to purchase 520 Bitcoins at a price of $67,068, increasing its total holdings to 847,363 Bitcoins.The Japanese listed company Metaplanet did not purchase any Bitcoin last week, marking nine consecutive weeks without purchases.Additionally, four other companies purchased Bitcoin last week. The Japanese food brand DayDayCook announced on June 17 that it spent $7.43 million to purchase 95 Bitcoins at an undisclosed price, bringing its total holdings to 2,899 Bitcoins; the Brazilian Bitcoin company OrangeBTC announced on June 21 that it invested $1.15 million to purchase 18 Bitcoins at a price of $64,121, increasing its total holdings to 3,822 Bitcoins; the asset management company Strive announced from June 15 to June 21 that it spent $49.98 million to purchase 759 Bitcoins at a price of $65,850, raising its total holdings to 19,864 Bitcoins.As of the time of writing, the total amount of Bitcoin held by the global listed companies (excluding mining companies) in the statistics is 1,142,276 Bitcoins, an increase of 1.87% compared to last week, with a current market value of approximately $7.417 billion, accounting for 5.7% of the circulating market value of Bitcoin.

According to Andrew Curran, the next-generation Mythos model from Anthropic, which has stronger capabilities (named either Mythos 5.1 or Mythos 6), has completed training. It is currently uncertain whether this model will be publicly released or kept for internal use only to accelerate the development of subsequent technologies.Analysis indicates that although some cutting-edge models like Fable 5 or Mythos 5 face restrictions or scrutiny regarding public release, this has not slowed down the research and development pace of top AI laboratories. In the face of fierce competition from open-source models represented by GLM-5.2, leading AI companies still need to continuously invest in and train more powerful systems to maintain their commercial and technological leadership.