Taiwan’s Ministry of Digital Affairs, through its Administration for Digital Industries, is nearing the end of applications for its 115 annual free AI computing program, with the final deadline set for Oct. 30. Approved domestic companies can access GPU resources on the government-backed platform free of charge for three months, and may apply more than once depending on their needs. The program supports both model training and inference workloads, allowing applicants to fine-tune open-source models with their own industry data or connect to preinstalled large language models through APIs for application development. According to the agency, by the end of year 114, 186 AI startups and information service providers had used the platform, producing at least 266 models or innovative applications. Remaining application windows now cover two training batches and three inference batches. The platform also includes a mix of local and international open-source models, including TAIDE, FoxBrain, Llama, Phi, Magistral-Small, Gemma 4, and openai/gpt-oss-120b, alongside tools for image and speech recognition.

Jiahui Yu, a core researcher in Meta’s superintelligence effort, has announced that he is leaving the company to start a new venture. Yu joined Meta in late June 2025 and disclosed his departure on August 14, 2026, ending a stint of a little over one year. His personal website has already been updated to reflect the change. The move comes only days after Meta released Muse Spark 1.2, the latest update in a product line developed by the team Yu helped build. In his public statement, Yu said working with CEO Mark Zuckerberg and Meta Chief AI Officer Alexandr Wang to build TBD Lab had been 「inspiring and fulfilling」. He also referenced a question he described as highly important to humanity yet still underexplored, saying it would now receive his full attention. He did not disclose the name of the new company, its research agenda, or any co-founders, adding only that there would be more to share as the work takes shape. Yu was one of the most watched names in Meta’s AI hiring push, during which reports and public comments fueled speculation around compensation packages at the $100 million level. While Meta executives later said such figures did not apply broadly and could reflect stock, bonuses, tenure, and performance conditions, Yu remained closely associated with that recruitment wave. Before Meta, he worked at Google Brain and OpenAI and contributed to projects including Gemini, GPT-4o, and GPT-4.1.

Meta has reopened its open-weight AI strategy with the release of Muse Glimmer, a model whose underlying parameters can be downloaded and modified by developers. The company said it will also release the weights for the more capable Muse Spark 1.2 in the coming weeks, marking a clear shift back toward the approach it once used to distinguish itself from rivals. CEO Mark Zuckerberg backed the move in a post on Meta’s website, arguing that powerful and free AI should reach billions of people rather than remain concentrated in large institutions. He also defended model distillation and said people should retain the ability to learn from observable information, a position that cuts against recent complaints from OpenAI and Anthropic over how their closed models’ outputs are used. Muse Glimmer is Meta’s first open-weight model since Llama 4 and its first to ship under the Apache 2.0 license. The model supports text and image input, carries a 128K context window, and is aimed at local agent use on personal devices. Meta’s release also comes with a broader commercial and infrastructure angle, as the company plans to spend as much as $145 billion on AI infrastructure this year while building out related cloud services.

Meta is shifting how it wants to compete in AI, according to a roughly 6,500-word open letter from Mark Zuckerberg titled The Future is for Everyone. The core pitch is that superintelligence should not sit in the hands of a small number of companies and institutions, but become a tool available to ordinary users. In practice, that means Meta plans to offer a free baseline AI product to billions of people, bring back some open-weight releases, and charge for heavier usage through a dynamic bidding system for additional compute. The article argues that this is more than a product message. It is also a competitive repositioning. If the contest is defined only by who has the strongest frontier model, Meta does not appear to hold a clear edge. But the company still controls major consumer distribution through Facebook, Instagram, WhatsApp, and a newer hardware entry point in AI glasses. On that view, Meta can afford to give ground at the model layer if it retains the user relationship and the right to price scarce compute. The write-up also points to Meta’s release of the 30-billion-parameter Muse Glimmer, its plan to open the weights for Muse Spark 1.2, and a $1 billion Future is for Everyone Fund tied to communities hosting data centers.

DuckDuckGo reported a sharp rise in US app installs and AI-free search traffic after Google unveiled a major AI search overhaul. The company said the issue is not AI itself, but whether users still have an option to turn it off.

Google has delayed the release of Gemini 3.5 Pro by months after the model failed to meet internal performance goals, especially in AI coding, according to a Bloomberg report cited in the source material. The model, internally known as “Cappuccino,” had been widely discussed online in the past 48 hours, with leaks pointing to a 2 million-token context window and a new “Deep Think” reasoning mode. Those expectations abruptly reversed after the report said Google had updated training data late last month in an attempt to improve coding performance, only to see disappointing results. The delay quickly spilled into the market. Google shares fell as much as 4.43% after the news. The report also described broader internal problems: complex management layers, competing priorities across major product lines such as Search, Maps, and YouTube, repeated overlap between teams, and limited compute access for engineers using internal AI tools. That tension stands out against the company’s projected 2026 capital expenditure of $180 billion to $190 billion and first-quarter capex of $35.7 billion, more than double a year earlier. The episode has also fed a wider debate about whether frontier AI labs are running into a new scaling wall. Ethan Mollick of Wharton described the pattern as a “next-generation giant-model disappointment trap,” linking Google’s setback to similar concerns around Meta Llama 4 and xAI Grok 4.

SemiAnalysis argues that Meta could move past Google within the next six months and become the strongest challenger behind OpenAI and Anthropic, though it says that outcome is still far from certain. The report rests on three main pillars: Meta’s $14.3 billion investment in Scale AI and the addition of founder Alexandr Wang, a reported shift of about 3,000 engineers toward building reinforcement learning tasks, and multi-gigawatt computing expansion tied to the company’s superintelligence push. The report does not claim Meta is already at the frontier. Meta released Muse Spark in April, and Axios reported on July 9 that Muse Spark 1.1 had opened its API to developers at $1.25 per million input tokens and $4.25 per million output tokens. Axios also said the model was not the leap Meta had hoped for, while a larger model codenamed Watermelon remains in training. SemiAnalysis frames the debate around speed of improvement rather than current rankings. It says Meta has redirected money, talent, engineering capacity and data center resources into its superintelligence lab after Llama 4 stumbled. Even so, the key question remains the same: whether Meta’s next-generation models can translate hiring, RL data pipelines and infrastructure buildout into products that actually narrow the gap with the leading labs.

Meta’s Llama project has seen 78% of its original research team depart, with many joining or founding rival AI startups. The exits, combined with leadership changes and product delays, are raising fresh questions about Meta’s long-term position in open-source AI.