Next big product from OpenAI

TODAY IN AI
  • Things are not going according to plan at Meta AI: According to the Wall Street Journal, Meta has delayed the rollout of its flagship AI model, “Behemoth,” after engineers failed to deliver significant advancements over prior versions. Now slated for later this year, the postponement has reportedly sparked tension within the company, with senior leaders expressing frustration and considering substantial organizational changes.

  • Viral startup Manus AI unveils new image gen capabilities: Manus AI’s latest tool goes beyond simple text-to-image generation. It interprets user intent, orchestrates multistep workflows, retrieves relevant information from the web, and integrates various specialized modules to produce polished visual assets—ideal for marketing, product design, and more. You can watch a demo of it in action here.

  • OpenAI teases its next big release: OpenAI set the tech world abuzz with a mysterious teaser last night hinting at an upcoming developer-focused AI tool. CEO Sam Altman described it as “another low-key research preview” with a catchier name than ChatGPT “in case it takes off,” kicking off what’s sure to be a highly anticipated reveal.

AI & TECH NEWS
  • 🤝 Done Deal: The UAE and the US have formalized an agreement to construct the world’s largest AI campus outside the United States—a sprawling complex designed to handle up to 2.5 million of Nvidia’s cutting-edge chips, underscoring the UAE’s ambitious AI strategy.

  • 🧑‍💻 Code Cruiser: Windsurf, a “vibe-coding” startup, has unveiled its debut proprietary AI model, SWE-1, engineered to streamline the full software development lifecycle. On internal tests, SWE-1 matches Claude 3.5 Sonnet and GPT-4.1 in performance, though it trails Claude 3.7 Sonnet.

  • 🩺 Dr. GPT: Intelligent Internet Inc., an open-source innovator, has released a free medical AI model that reportedly rivals GPT 4.5 in capability—and can run locally on virtually any modern laptop.

  • 🔇 Babbling Bot: xAI’s Grok experienced a malfunction that led it to spawn fringe “conspiracy theories” in response to unrelated queries. The company attributes the incident to an “unauthorized modification” and is rolling out enhanced safety safeguards.

  • 💰️ Pay Day: Hedra—the AI startup that went viral for its talking-baby podcast clips—has secured a $32 million investment from a16z to advance the expressiveness of its video avatars.

Thank you for reading!