Meta's Watermelon Model, Cloudflare's AI Bot Controls, and New Funding Rounds
Meta's new 'Watermelon' model matches GPT-5.5 performance, Cloudflare introduces granular AI traffic controls, and major funding rounds hit Kling AI, Symbotic, and Tripo AI.
Podcast В· 3 min
Meta Teases 'Watermelon' Model Matching GPT-5.5
Meta's superintelligence chief Alexandr Wang has reported that the company's new model, 'Watermelon,' currently in training, has achieved performance parity with GPT-5.5. The model is reportedly running on approximately 10 times the compute resources of its predecessor, Muse Spark, which launched in April. While CEO Mark Zuckerberg noted at a recent town hall that agent development has progressed slower than expected, Wang clarified that the company is preparing to release an Opus-level coding model soon, alongside updates to Muse Spark that will bring significant improvements to coding and agentic capabilities. This development signals Meta's continued aggressive investment in its AI stack, aiming to close the gap with frontier leaders.
Cloudflare Introduces Granular AI Traffic Controls
Cloudflare has launched a new set of AI traffic controls, allowing website owners to distinguish between different types of bot traffic, including search, agent, and training bots. Starting September 15, 2026, new domains on the platform will block agent and training bots by default on ad-supported pages, while allowing search indexing. This move aims to provide publishers with a clearer permissions system for their content, addressing the growing tension between site owners and AI companies that use web data for model training. The update effectively turns AI crawling management into a standard business decision for online publishers.
Amazon Designs Custom AI Chips for Devices
Amazon has announced it is designing custom AI chips specifically for its Echo, Fire TV, and future hardware devices. This initiative aligns with the broader industry trend of major tech companies seeking to gain greater control over their hardware and software stacks by reducing reliance on third-party silicon providers. By developing specialized chips, Amazon aims to enhance the performance and efficiency of on-device AI capabilities across its product ecosystem.
Zoom Acquires Common Room
Zoom has agreed to acquire Common Room, a move designed to integrate buyer-signal intelligence into its AI revenue platform. The acquisition will incorporate Common Room's technology and 'RoomieAI' agents into the Zoom Revenue Accelerator, aiming to provide sales teams with deeper insights into customer intent and engagement. This strategic addition is part of Zoom's ongoing effort to expand its enterprise AI offerings beyond communication tools.
Kling AI, Symbotic, and Tripo AI Secure Funding
The AI sector continues to see significant capital inflows. Kling AI, a video AI unit spun off from Kuaishou, reportedly raised an initial $2 billion. Warehouse robotics firm Symbotic acquired ARMS Innovations to advance AI-powered warehouse orchestration. Additionally, Tripo AI secured $150 million to further develop its 3D foundation models and world-model tools, highlighting continued investor interest in specialized AI applications and infrastructure.
Lenovo Launches $44 AI Phone for Students
Lenovo has launched an 'AI Student Phone' in China, priced at approximately 299 yuan ($44 USD). The device is designed with a focus on basic functionality, featuring a small screen, parental tracking, and a dedicated AI button for homework assistance. It excludes web browsers and social media, aiming to provide a distraction-free tool for students. The phone includes features such as SOS dialing, GPS location tracking for parents, and scheduled usage hours, positioning it as a specialized, safer alternative to standard smartphones for younger users.