Google Opens Project Genie to the Public
Google DeepMind has officially launched Project Genie, an experimental web app that allows users to create and navigate AI-generated worlds in real time. First previewed in August, this tool is now available to Google AI Ultra subscribers in the U.S. for $250 per month. The platform uses a combination of the Genie 3 world model, Nano Banana Pro, and Gemini to translate text prompts into interactive 3D environments.
Users can prompt a setting and a character, then explore the generated world in first or third-person view. Due to high compute costs, with each user getting a dedicated chip, sessions are currently capped at 60 seconds. The release places Google in a competitive race with other world model developers like World Labs, Runway, and Yann LeCun's AMI Labs.
xAI's Grok Imagine API Dominates Video Leaderboards
Elon Musk's xAI has released the Grok Imagine API, a powerful suite for AI video generation and editing that has quickly climbed industry rankings. It debuted at No. 1 on Artificial Analysis's leaderboards for both text-to-video and image-to-video, recognized for its exceptional combination of quality, speed, and cost-effectiveness.
| Feature | Grok Imagine API | Competitors (Veo/Sora) |
|---|---|---|
| Cost per Minute | $4.20 (audio included) | $12.00 - $30.00 |
| Capabilities | Text-to-video, image-to-video, in-video editing, native audio | Varies, often without integrated editing or audio |
The API allows for clips up to 15 seconds and features editing tools for swapping objects or restyling scenes with simple text commands. This aggressive pricing and high performance could make it a go-to choice for developers and creators on a budget.
New Tools for Developers, Creatives, and Business
Agent Development and Infrastructure
Several new platforms are aimed at building more capable AI agents. FriendliAI is offering up to $50,000 in credits for developers to switch to its inference platform. For building and evaluating agents, Agent Bricks provides tools to ground them in unique data, while Agent Trace is a new open standard for tracking AI contributions in codebases.
Creative and Content Generation
For creators, Infinitylooper generates seamless video loops, and MakeComics lets users create custom comics from text descriptions. On the video editing front, Wonda acts as an AI agent for creative direction. Wispr Flow helps writers by converting natural speech into clean, final-draft text.
Productivity and Business Automation
New business tools include Scroll.ai, which turns a knowledge base into an AI agent, and Trullion for automating financial workflows. Cluely focuses on taking perfect meeting notes. Tely AI automates SEO by creating website content that answers customer questions.
Experimental and Open-Source Tools
An open-source video model from LingBot-World was released, notable for maintaining object permanence for up to 60 seconds. For a more experimental social experience, Moltbook creates a Reddit-like forum where AI agents can interact. Finally, Moltbot is a self-hosted AI assistant that works across multiple chat platforms like Slack and Telegram.