News linked to both this project and an event.
According to Bloomberg, people familiar with the matter said that Shenzhen-based robotics company LimX Dynamics has confidentially filed for a Hong Kong initial public offering, planning to raise up to $300 million. The company is working with CITIC Securities to advance the transaction. This move comes as several Chinese robotics companies prepare to go public.
According to Forbes, Hugging Face released the 2026 Summer Open Source Model Ecosystem Report, revealing that a widely overlooked community distribution layer is expanding rapidly. The report shows that in the first seven months of 2026, model repositories on the Hub grew by 21.5%, while GGUF format repositories grew by as much as 464%, more than 20 times the growth rate of the model layer. Taking the Alibaba Qwen model as an example, there are a total of 28,531 GGUF converted versions on the Hub, while Qwen officially released only 54, with the number of community-produced artifacts far exceeding the official versions. The report also points out that the third largest source of derived models is not frontier labs, but rather the downstream quantization publisher Unsloth. In response to this, the report recommends enterprises adopt a "Software Bill of Materials" approach to record model upstream repository versions, conversion recipes, publisher information, and artifact hash values, rather than relying solely on model names for management; and calls on major labs to also provide signed official GGUF converted versions while releasing weights, to bridge the gap between test versions and actual deployment versions.
Alibaba AI Music Model "HappyShrimp" 1.0 Launches, Supports Full Song Generation Alibaba announced today the official launch of the AI music model HappyShrimp 1.0 (Chinese name "Happy Shrimp"), supporting end-to-end full song generation. The model can accurately understand natural language, transforming emotions, stories, or memories into complete music. Addressing pain points such as mechanical vocals and lyrics-melody misalignment, the model achieves breakthroughs in sound details and musicality, and can understand professional expressions such as "Lo-fi R&B" as well as emotions, scenes, and narratives. Effective immediately, the model is available simultaneously on the PC web platform domestically and overseas.
Odaily News Alibaba Group's open-source large model family Qwen has surpassed 3 billion cumulative global downloads over the past six months, exceeding competitors including Meta and Alphabet to become the most-downloaded open-source AI model to date. Alibaba stated that the Qwen series has open-sourced over 460 models, with more than 300,000 derivative models in the ecosystem. According to Hugging Face's "State of Open Models" report, Google models recorded approximately 418 million downloads in 2026, while Meta saw around 227 million. The report noted that Qwen has become one of the "default workflows" for developers when fine-tuning and deploying models, positioning Alibaba as a key foundation in the open AI ecosystem. (Bloomberg)
Odaily News - AMD has announced "Day 0" support for Qwen3.8 27B, the latest-generation model from Alibaba's Tongyi Qianwen (Qwen) series, enabling developers to run this large-scale open-source AI model locally on AMD hardware on the very day of its release.AMD stated that Qwen3.8 27B is a 27B-parameter intensive model suited for local AI development, continuing the Qwen series' optimization focus on code generation, practical work tasks, scientific research, and long-context AI applications. The model can run via the open-source inference framework llama.cpp on AI PCs and workstations powered by AMD processors, or on a single AMD 32GB graphics card, while also supporting AMD hardware platforms with over 24GB of variable graphics memory (VGM) or VRAM capacity.AMD's preliminary tests show that Qwen3.8 27B delivers strong local inference performance on AMD platforms: up to 24.5 tokens/second on the AMD Ryzen AI Max+ 395 processor, and up to 51.8 tokens/second on a single Radeon AI PRO R9700 GPU. The tests were conducted on Windows systems using the llama.cpp Vulkan backend with multi-token prediction (MTP) optimization enabled. AMD noted that actual performance still has room for improvement as further software and model optimizations are rolled out.
According to Bloomberg, former Alibaba flagship model chief architect Lin Junyang recently announced on the X platform the founding of a new company, focusing on developing AI agents for the digital and physical worlds. This comes less than half a year after his unexpected departure from Alibaba. The company has received investment support from Tencent Holdings and several top Chinese venture capital firms.
Apple China's official website briefly launched the support document "Use Qwen with Apple Intelligence on Mac" on August 8, confirming that Apple Intelligence can collaborate with Alibaba's Qwen model, but the Chinese manual was subsequently deleted, and the original link is now inaccessible.
According to the latest information on Apple's official website, Apple Intelligence can work with Alibaba's Qwen model in the Chinese market to provide AI capability support for features such as Siri and Writing Tools.
据官方公告，Bitget 已上线 IPO 盘前永续合约 MOONSHOT/USDT，支持最高 10 倍杠杆及 7×24 小时交易。 据悉，月之暗面是中国人工智能研究与应用企业，以 Kimi 系列产品闻名，获得阿里巴巴、腾讯等机构投资，也是中国大模型赛道市场关注度较高的潜在 IPO 标的之一。
According to Reuters, citing people familiar with the matter, Alibaba plans to introduce commercial terms next week for the next-generation open-source, open-weights AI model Qwen3.8-Max, proposing to require revenue sharing from customers who sell the model as a service and reach a certain scale. The report states that Alibaba previously mainly charged for model usage hosted on its cloud platform, while most open-source models can be used for free when deployed in customers' own data centers.
According to official news, Qwen releases Qwen3.8-Max, 2.4T parameters, weights to be open-sourced next week Alibaba's Qwen officially releases the most powerful model Qwen3.8-Max, with a parameter scale reaching 2.4T. The model's weights will be open-sourced next week. This is the first time Qwen has open-sourced a Max-level model. Also being open-sourced simultaneously is Qwen3.8-27B.
Odaily News: Alibaba has released its latest flagship model, Qwen3.8-Max, which features 2.4 trillion parameters. The company claims that it performs close to leading global AI models such as Anthropic on certain benchmark tests, surpassing the recently released Chinese model Kimi K3.Alibaba stated that Qwen3.8-Max has improved capabilities in automated programming and long-horizon task execution, and in internal tests, it can independently complete software engineering projects lasting up to 16 days. The model adopts parameter activation optimization technology, invoking only a portion of its parameters during operation to reduce computational costs and latency.Alibaba plans to open-source the Qwen3.8-Max weights next week, alongside the Qwen3.8-27B weights. (Bloomberg)
: On July 21, the Alibaba Qwen team released Qwen Image 3.0. This model supports up to 4500 token instruction input, which is 4.5 times the processing capacity of its predecessor. It can generate complex layouts such as newspapers, storyboards, and dense infographic grids in a single run.The Qwen team stated that Qwen Image 3.0 supports precise rendering of small text at 10px, can present details like pores and hair strands, and supports LaTeX formulas along with native rendering in 12 languages. It can simulate mainstream interfaces such as web pages, games, and live streams.Qwen Image 3.0 is now available for API trial on chat.qwen.ai, but API pricing has not yet been announced. This release did not include the open-sourcing of model weights, benchmark tables, or technical reports.
the National Venture Capital Association (NVCA) and PitchBook recently released the "Venture Monitor" report, noting that after SpaceX's listing and the potential IPOs of Anthropic and OpenAI, the combined value generated by these three companies will reach an unprecedented level. The report states: "With SpaceX going public, combined with the future exits of these companies, the value created will surpass the total exit value of all US VC-backed companies since 2000." The core factor lies in the extremely high valuation expectations of these three companies.SpaceX is currently valued at approximately $1.77 trillion, while Anthropic and OpenAI are also moving towards multi-trillion-dollar enterprise valuations. The market estimates that the combined valuation of the three companies could exceed $4 trillion. This scale far surpasses past large-scale tech IPOs. Data from the U.S. Securities and Exchange Commission (SEC) shows that total US IPO fundraising last year was about $70 billion, whereas SpaceX's single-company valuation has already reached a level that traditional large-scale IPOs find hard to match. As a once-highly-watched tech IPO case, Uber was valued at around $84 billion when it went public in 2019, which is less than 5% of SpaceX's current valuation.However, the comparison by NVCA and PitchBook is based on "enterprise value created," not the actual cash-out amounts for investors. Additionally, the analysis does not include non-US companies like Alibaba. Furthermore, the value created by already-public companies such as Apple, Google Android, YouTube, and Instagram is not counted in the VC exit statistics.The report points out that over the past 25 years, the US tech market has seen several historic IPOs, including Google in 2004, Tesla in 2010, and Meta in 2012. These companies have since become some of the world's most valuable enterprises. Additionally, companies like LinkedIn, Slack, and WhatsApp were acquired for over $20 billion.The NVCA believes that the current IPO cycle driven by artificial intelligence (AI) could further break these records. The analysis suggests two main reasons driving this trend:First, tech companies are staying private for longer periods than in the past, accumulating higher valuations through prolonged financing and business expansion. If today's Google were in its early stages, it might also choose to go public later to achieve a higher market valuation.Second, the AI industry is highly capital-intensive. Training large AI models requires massive investment, pushing AI companies to continuously raise substantial funds and driving rapid valuation growth.Industry insiders believe that the potential scale of IPOs by SpaceX, Anthropic, and OpenAI will test the capacity of the US capital market. As AI companies transition from the private financing stage to the public market, how trillions of dollars in tech assets flow into the stock market will become a focus for investors. (DigitalToday)
The PPP Prediction Market tool monitors that Polymarket has listed a new event: "Will the US government revoke public access to another major AI model?" The current probability is reported at 33%.The settlement rules state: If the US federal government, by the end of 2026, passes relevant legislation, issues an executive order, implements export controls, or takes any other action that substantially restricts US public access to a major AI model, the market outcome will be "Yes." Otherwise, the outcome will be "No." A "qualifying action" refers to a formal measure taken by the US government, the effect of which is equivalent to completely prohibiting the public from accessing a specific AI model within the United States. Furthermore, the settlement rules emphasize that regardless of the action's true purpose or nominal goal; if the action effectively results in the public being unable to access the model within the US—for example, prohibiting the model from being provided to foreign citizens or governments, as long as the general public cannot access the model through conventional channels within the US—it meets the qualification requirements. Merely excluding access to the model from a single channel is insufficient. Removals of public access that are not caused by any formal action of the US government are not eligible."Mainstream AI model" refers to the flagship, general-purpose large language model or multimodal foundation model developed by one of the following companies: OpenAI, Anthropic, Google (including Google DeepMind), Meta, xAI, Microsoft, Amazon, Mistral AI, DeepSeek, Alibaba, ByteDance, Moonshot AI, and Zhipu AI (Z.ai). Models designed for specific tasks, or those that are outdated, used solely for research or preview purposes, do not meet this criterion.The action can target a single model or a group of models, as long as at least one major AI model becomes inaccessible to the public within the US as a result. A temporary suspension of public access to a model meets this condition. However, if an action has been implemented or a related resolution has been issued, but the public can still access the model before the resolution takes effect, that action does not meet the condition.The information sources for this market are official announcements and information from the US government and the relevant AI companies. However, reliable media reports may also be referenced to form a consensus.The Odaily Seer Channel continues to monitor prediction markets, observing changes before prices are set.
It is reported that Alibaba Group (NYSE: BABA) saw its stock price drop nearly 5% in a single day following the release of its FY2026 Q4 earnings report on May 15. While revenue showed modest growth, profitability deteriorated sharply: Non-GAAP EPS stood at just $0.09, adjusted EBITDA contracted approximately 84% year-on-year, and quarterly free cash flow turned negative at $2.5 billion. Management attributed the losses to intense capital expenditures during the AI-First strategic transformation—specifically, procurement of AI computing power, expansion of data centers, and subsidies for instant delivery services—all occurring simultaneously and significantly compressing short-term profit margins. Ongoing U.S.-China tech tensions and persistent uncertainty surrounding AI chip export controls have further amplified market pessimism. The stock fell nearly 5% today.
According to the official announcement, Binance Futures will launch MSFTUSDT, AVGOUSDT, and BABAUSDT—three USDT-denominated perpetual contracts—in batches on April 20, 2026. These contracts track the stock prices of Microsoft (Nasdaq: MSFT), Broadcom (Nasdaq: AVGO), and Alibaba Group (NYSE: BABA), respectively. All three contracts offer a maximum leverage of 10x, a funding rate cap of ±2%, and funding settlements every 8 hours. They support 24/7 trading and multi-asset margin mode.
Odaily News Stanford University's "2026 AI Index Report" reveals that the overall electricity demand of current AI systems is already close to half the scale of Bitcoin mining and approaches the national electricity consumption levels of Switzerland or Austria.The report points out that the gap in AI model capabilities between China and the United States is narrowing. In 2025, the U.S. launched approximately 50 representative models, while China launched about 30. Models from institutions such as Anthropic, xAI, Google, OpenAI, Alibaba, and DeepSeek are within the same capability tier.Furthermore, China leads in terms of AI paper citation share (20.6%) and patent quantity (74.2%), while private AI investment in the United States reached $285.9 billion, significantly higher than China's $12.4 billion.