Google DeepMind releases Genie 4 world model
Google DeepMind: Genie 4: A new generation of world simulation
Genie 4: our new world model unlocking unprecedented realism in real-time simulation.
What happened
Google DeepMind released Genie 4, a world model that generates photorealistic interactive simulated environments in real-time based on text prompts, with real-time audio and the ability to edit scenes mid-simulation. The system improves on its predecessor with more realistic physics, better human movement, longer scene consistency of up to two minutes, and closer integration with Google's Gemini AI model. The technology is currently in research preview for selected testers, with broader access planned through an API for developers and through subscription tiers for consumers.
What each outlet adds
- Google DeepMind ↗: The system generates visuals, audio, and character behavior simultaneously and is described as crucial for building artificial general intelligence.
- TechCrunch ↗: World models are being developed for commercial applications including gaming, interactive learning, robotics, film prototyping, and testing robot scenarios before real-world deployment.
- Ars Technica ↗: Access will roll out to developers through an API and to consumers through the most expensive subscription tier, with pricing not yet announced.
Why it matters
World models are considered important for developing artificial general intelligence and could provide training environments for robots or enable rapid prototyping in creative fields. The release comes as the world model industry consolidates, with several companies and startups acquiring or being acquired to develop simulation capabilities.
Get smart on it
Genie 4 is a world model that can simulate realistic interactive environments in real-time, generating visuals, audio, and character behavior simultaneously based on text prompts. The system builds on existing multimodal AI capabilities to maintain coherence and continuity as users modify scenes through natural language commands. Creating simulations of this kind matters because understanding and simulating the world is described as a crucial aspect of building artificial general intelligence, with potential applications ranging from robot training environments to new interactive experiences. The technology is currently in research preview for selected testers, with broader access planned for certain subscriber groups.
Relevant courses
Want to go deeper than the news? Explore live, cohort-based AI courses taught by practitioners.
Browse AI courses on MavenRelated news
Introducing GPT-6.1 Terra
Bringing fast and capable computer use to everyone
Also reported by SiliconANGLE, MarkTechPost and The Decoder
Google’s next Genie AI model generates more realistic experiences
Google announced today its Genie 4 world model for generating interactive experiences using AI. The company says the model has four main updates: It allows for more realistic worlds, lets users update an experience in real time with a natural language prompt, can generate audio, and uses Gemini 3.8 Flash under the hood to help with its simulations. It also can create experiences that are about two minutes long, up from one minute for the previous Genie 3 model. Google is pitchi
Moonshot AI Releases Kimi K3.1 Pro: A 2.8T Total/104B Active Params (MoE) Model with 1M Context
Moonshot AI has just released Kimi K3.1 Pro, the first model in its K3.1 series. Kimi K3.1 Pro is a 3T-class open-weight* model with a 1-million-token context window. Moonshot team positions it against the strongest proprietary models on coding, agentic and knowledge benchmarks. For developers, it is cheaper API pricing than K3 and open weights yet not available. *open weights are expected to release in November 2026. TL;DR Size: 2.8T total, 104B active parameters (MoE), 1M-token context
Also reported by SiliconANGLE and The Decoder
Mirendil Launches Fern 0.1, Its First Model, With 10M-Token Context
Mirendil has released Fern, its first AI model. Mirendil Fern is a from-scratch coding model with a context window of up to 10M tokens. Mirendil explains how that equals about 23 MB of code, or roughly half a million lines. That is a codebase on the scale of Megatron-LM, held in context at once. Fern 0.1 is the first version, built from Mirendil-Base after a short round of post-training. For developers, its like, you load a whole repo, a full training log or days of agent traces at once. TL;
Jack Dorsey’s Block made an AI model that talks to you like an awkward new friend
Jack Dorsey's Block announced an AI model called Dessa that's trained "from scratch on nothing but human conversation," and to show off the model, Block shared chats it has had with humans that can occasionally be a little awkward and uncanny. Take this one, for example, shared by Dessa lead Ragavan Thurairatnam. The conversation between the model and the speaker does have a mostly natural back and forth (though for some reason, the model initially mistakes the speaker for some
Also reported by TechCrunch and The Decoder
Canva releases its first image generation model
Canva is making its new image model available to all users, promising better, more customizable performance at lower costs.
Also reported by The Verge and The Decoder