Likely boosted by Donald Trump's mention of Super Intelligence, Solana ecosystem meme coin Super Inu surged more than 3 times in a short timeframe.

Viking Therapeutics Inc. shares rose sharply after the company announced positive results from a small exploratory study of its dual-target injectable obesity drug, which helped patients achieve weight loss and maintain the effect even when dosing frequency was reduced. The company drew conclusions based on data from the study’s extended patient cohort. The research was originally designed to determine the optimal dose of the drug, which regulates weight through two distinct pathways. The results are a major boost for Viking, which had faced setbacks in prior trials. The company said patients on a moderate weekly dose in the main trial phase saw the best outcomes, shedding up to 19% of their body weight over roughly five months. A 13-patient "exploratory" group continued on a slightly lower dose for an additional three months, achieving 22% weight loss with no plateau observed. Viking noted that when patients switched to a less frequent dosing regimen (once or twice monthly), some were able to maintain 90% to 97% of their weight loss. Side effects were mostly mild and eased over time, with only one patient discontinuing treatment due to adverse events. During low-frequency maintenance dosing, gastrointestinal side effects were comparable to those in the placebo group. Viking’s stock jumped as much as 36.6% during New York trading, marking its largest intraday gain since July 2024, before paring some of those gains.

According to market data from BIT (bit.com), the declines of major US bank stocks have widened: JPMorgan Chase fell 4.1%, Bank of America dropped 3.1%, Wells Fargo declined 4.2%, Citigroup fell 3.2%, Goldman Sachs dropped 1.9%, and Morgan Stanley declined 3.6%.

According to official news, PayPal plans to partner with Meta to use the Muse proxy for shopping and checkout.

Beating AI News and LatePost report that Alibaba Cloud’s无影 team is developing QwenBook, a native agent-focused computer for office use, designed in a form factor similar to a tablet paired with a keyboard. The product’s definition and hardware are developed in-house by the team, leveraging无影’s existing systems and end-cloud capabilities, and currently integrates Qwen3.8-Max and Qwen3.8-Flash models. IT Home also witnessed a live demo of the device at today’s Yunqi Conference. Simply put, Alibaba aims to embed Qwen directly into the system, enabling agents to access context such as files and meeting notes, then call tools to complete tasks—rather than functioning solely as a chat app. QwenBook already includes built-in applications like audio transcription, Wiki, smart folders, email, MyNote (knowledge base), and MyZone (memory management), and is collaborating with Kingsoft to integrate WPS. This approach differs from that of Doubao Phone, which primarily operates existing apps on the device via screen reading and simulated clicks; QwenBook, by contrast, starts with office applications it controls natively. Doubao has already faced limitations with third-party apps along this path. The newly launched nubia NaviX Ultra still cannot automatically execute tasks within apps including WeChat, Taobao, Meituan, and Xiaohongshu. QwenBook remains an early trial project. LatePost notes the team has around 150 members, launched the project around April this year, and the full product is expected to launch by the end of 2024 or early 2025. Alibaba’s current strategy is to first build an office environment where agents can actually perform tasks, before gradually integrating third-party software.

Beating AI News – Andreessen Horowitz (a16z) has invested $35 million to launch the Andreessen Academy, a program for young people aged 16 to 22. Ten tech companies including OpenAI, Anthropic, Google, Meta, NVIDIA, Palantir, and Stripe are its founding partners. This academy deviates significantly from traditional university models: admissions prioritize applicants’ existing achievements, and students spend up to 80% of their time on hands-on projects such as training AI models, building products, or launching startups. Each student will also receive over $50,000 worth of computing power and technical resources. Instructors are mostly from tech companies and the startup ecosystem, including figures like Sam Altman, Li Feifei, and Figma CEO Dylan Field. Students also gain direct access to learning and internships at partner companies; over 40 firms including Cognition, Databricks, Notion, and Shopify have joined as recruitment partners. a16z is also integrating startup funding into the program, with partner Erik Torenberg stating the firm will likely invest in entrepreneurs emerging from the academy. The academy unifies processes previously scattered across universities and various Silicon Valley sectors—from admissions and education to internships, recruitment, and startup funding. Its inaugural cohort will enroll around 50 students, with classes starting in fall 2027, and the first year is tuition-free. The official two-year program is set to launch in 2028, pending regulatory approval and without university accreditation for now. Future tuition is expected to align with that of top U.S. private universities.

Beating AI Express News: Cloudflare has launched Worker Previews. Workers can be simply understood as backend programs running on Cloudflare, used to handle APIs, logins, databases, and web requests. Now, every time a developer or Coding Agent creates a new Git branch, they can spin up an independent temporary test environment for these Workers. They can modify code and deploy it directly for testing without touching the production site. This test environment comes with its own URL, logs, and runtime status, and can also create separate resources such as Durable Objects and Containers. As a result, Agents can independently complete the full workflow: modifying code, deploying, opening web pages for testing, checking errors, and making further revisions. Cloudflare previously offered Preview URLs, but these mainly generated temporary URLs for a specific Worker version, with underlying resources still potentially linked to the production environment. Now each branch can run a separate instance, so Agents no longer have to worry about disrupting live services when testing new code.