els

Arthur Hayes is optimistic about Anthropic's initial public offering performance, stating that he is increasing his crypto risk positions and maintaining a small to medium size for quick liquidation when necessary.

According to the Science and Technology Innovation Board Daily, Doubao Work has officially integrated with multiple third-party models such as Kimi 3.1 Pro, GLM 5.3, and DeepSeek V4.1 Flash.In addition, Doubao Work has launched a limited-time discount for Doubao 2.1 Lite and DeepSeek V4.1 Flash, reducing the consumption multiplier for these two models.

MiniMax officially released H3-Image and open-sourced the model weights. This model can directly generate images of 4096×4096, approximately 16.8 million pixels, without additional upscaling. According to official data, the generation time is less than 10 seconds, with speeds reaching up to 5 times that of mainstream products. Text-to-image and image editing have been integrated into the same model, supporting up to 16 reference images.H3-Image follows the architecture of the H3 video model, removing the audio component while retaining the original text encoder and redesigning the decoder for images. The core image generation network has 33.1 billion parameters, requiring only 6 sampling steps to generate an image. On August 7, MiniMax researcher Kiro revealed in a Reddit AMA that the team is developing a dedicated image model based on the H3 homogenous architecture and plans to open-source the weights. They also proposed a workflow where the image model generates the first frame, which is then handed over to H3 for video generation.On November 3, Pruna AI launched P-H3-Image based on the open-source H3 video model, supporting raw images and retouching, compressing the generation network from 33.1 billion to approximately 20 billion parameters, requiring only 8 sampling steps. The minimal quantization version is about 20 GB, with peak memory usage of approximately 21 GB when generating 8 million pixel images. The official MiniMax version retains the 33.1 billion parameter generation network, with a sampling step count of 6, and supports higher resolutions and more reference images, with the original weights plus the reused text encoder totaling approximately 135 GB. There are regional restrictions on the open-source license; the United States, United Kingdom, European Union, and South Korea are not within the community license authorization regions. Companies with annual revenues exceeding 1 million dollars or those providing model hosting services to third parties must obtain separate authorization in advance.

On-chain data analysis platform Glassnode stated on platform X that the trading volume of altcoins has returned to normal levels. Throughout September, the total trading volume of altcoins had risen to four times that of Bitcoin, marking a period of active altcoin trading.Glassnode pointed out that this speculative altcoin activity often coincides with local market tops, when Bitcoin peaked around 87,000 USD.

The October Consumer Expectations Survey by the New York Federal Reserve shows that U.S. consumers' inflation expectations for one year, three years, and five years remain at 3.9%, 3.3%, and 3%, respectively. The average expected probability of an increase in the unemployment rate over the next year has risen to 45.3%, the highest since April 2020. The expected growth in household income has decreased to 3%, and the expected growth in consumer spending has decreased to 5.3%.

According to Catcher Predict monitoring, the "Sunderland vs Chelsea" event currently ranks high in trading activity on the prediction market Polymarket, with a cumulative pool size reaching 5.0753 million USD. The option with the highest market consensus is "Chelsea," corresponding to a winning probability of approximately 51.5%.

Federal Reserve's Williams stated that the U.S. labor market is exceptionally stable and robust, with remarkable economic resilience and enhanced growth. Currently, inflation levels in the U.S. are too high and have persisted for too long, and the inflation rate must be reduced to 2%.Energy prices affect ticket prices, and artificial intelligence has a limited impact on chip prices. The effects of tariffs have largely become a thing of the past.

According to a report by Yicai, the "2026 Corporate AI Status Report" released by Silicon Valley venture capital firm Menlo Ventures shows that 89% of the spending by American companies on open-source models went to Chinese models, with Kimi accounting for 61%, DeepSeek for 15%, and GLM for 11%.In addition, on the global AI model aggregation platform OpenRouter, from October 26 to November 1, the number of Token calls for Chinese large models has exceeded that of American models for 27 consecutive weeks, with Chinese models occupying four of the top five models in terms of platform call volume that week.

According to China Blue News, on November 6, at the 2026 World Internet Conference in Wuzhen, Sohu founder, chairman of the board, and CEO Zhang Chaoyang admitted that he "missed the mobile internet." In the face of major companies heavily investing in large models, he explained that the outside world perceives Sohu's attitude towards AI as "calm," but "it is actually a conscious choice," because "large models require resource competition, which we cannot afford."When discussing algorithms and social interaction, Zhang Chaoyang stated that he does not oppose algorithms, but in product design, he chose not to pursue algorithmic recommendations, believing that some users would prefer to obtain information through real human social interactions. Regarding the short drama sector, he judged that the popularity of short dramas in China has not been replicated in the U.S. market, where American users still prefer long dramas and movies, saying, "The phenomenon of short dramas may be fleeting, or it may be a part of the future." He revealed that he does not watch short dramas, let alone AI short dramas.

At the 9th Hongqiao International Economic Forum, Wang Xingxing, founder and chairman of the humanoid robot company Yushu Technology, stated at the sub-forum "Innovation and Development Cooperation of Humanoid Robots and Embodied Intelligence Industry" that the current core challenge of physical AI mainly lies in the alignment between models and the physical world.Wang Xingxing believes that the understanding capability of multimodal models is already quite good, and the biggest issue is the error in input-output alignment, including the deviation between sensor measurements at the input level and the actual state, as well as the deviation between action commands at the output level and actual movements. He predicts that this part of the problem may be solved in the next 1 to 2 years.

According to IT Home citing The Information, OpenAI may release the next generation of Astra and Sol series models, as well as the Cyber model aimed at the cybersecurity field, as early as November 6 (Friday) local time. The report did not disclose specific model numbers, pricing, context window lengths, or benchmark scores, nor is it clear whether these models will be launched first through API, ChatGPT, Codex, or partner channels.Previously, users had discovered the "gpt-6.2-astra" label in Codex usage statistics. Users suspected to be participating in the gray testing claimed that the new model has improvements in physical effects and animation generation, but these feedbacks are mainly based on subjective experiences, and there is currently no public benchmark data to verify.

According to "Du Jia," Baidu recently launched an open model service platform called "Word Element Institute" aimed at developers, enterprise clients, and model suppliers. It is positioned as a matchmaking trading platform connecting model suppliers and consumers, gathering 42 large models, covering text and video models such as DeepSeek, Kimi, and Zhipu GLM. Developers can connect to the Word Element Institute API once to switch between models from multiple vendors and be billed based on usage.The same model can be accessed by multiple suppliers. The platform displays pricing, initial token latency, throughput speed, 7-day success rate, and historical call volume, and supports both OpenAI and Anthropic protocols. Taking DeepSeek-v4.1-flash as an example, the platform connects to 6 suppliers, marking the price savings ratio on the page, while also displaying context window, launch time, deployment region, and supplier credit evaluation.The platform ranking is based on real call data and can be viewed for the last 24 hours, last 7 days, and last 30 days. Screenshots show a total of 398 million calls in the last 7 days, with a total token volume of 51.336 trillion, involving 68 ranked model resources, and an average initial token latency of 290 milliseconds. DeepSeek-v4.1-flash ranks first in both call volume and token usage. Participating suppliers include Bochuan Technology, Infinite Mind, Star Hub Word Element, and Zhipu Original Factory, with the platform showcasing resources and supplier information through credit evaluations and blockchain proof.

According to a report by Reuters on November 4, citing the Financial Times, Oracle has canceled a multi-billion dollar chip leasing contract with Tencent. The report stated that U.S. officials had expressed concerns to Oracle about the deal, and Oracle abandoned the agreement under pressure from the Trump administration.It was reported that Tencent also hosted high-level officials from Guangdong Province, who requested the company to support domestic chips. Reuters stated that it was unable to immediately verify the above report, and Oracle, Tencent, and the White House did not respond to Reuters' request for comment.The Financial Times reported last month that Tencent signed its largest overseas leasing agreement with Oracle, valued at approximately $7 billion. The five-year agreement covered multiple data centers of Oracle in Southeast Asia and was originally expected to allow Tencent to obtain about 100,000 advanced AI chips that could not be acquired in China due to related restrictions between Washington and Beijing.

The IPO dedicated website of AI company Anthropic, ipo.anthropic.com, is now live. The page is very simple, featuring three circles of increasing size, along with the phrase "More information to come." Developers examining the webpage code discovered that the site has preset paths for regions including the United States, the United Kingdom, Canada, Australia, and Japan, and also contains Japanese prompt text.Anthropic plans to issue stocks to retail investors in Japan, the United Kingdom, Canada, Australia, and some European countries, with participating institutions such as Interactive Brokers, Revolut, and SBI already announced. Japanese regulatory documents indicate that the local issuance scale is expected to be up to approximately $3.5 billion. The company publicly submitted its IPO prospectus on October 26, intending to list on NASDAQ under the code ANT, and may initiate the listing promotion during the week of November 9, aiming to complete the listing by the end of the month.