Introducing Genie 4

Raia Hadsell, Diego Rivas, and Hang Qi from Google DeepMind join host Logan Kilpatrick to introduce Genie 4, the latest world model from Google DeepMind. Their conversation covers: *What world models are *The arc from Genie 1 to Genie 4 *Live demos showcasing the model’s biggest leaps forward *The role of Gemini in the loop *The path to robotics and education applications *What it means to make Genie available in the API for the first time Chapters 0:00 — Welcome and Genie 4 introduction 1:02 — What are world models? 2:10 — The arc from Genie 1 to Genie 4 5:10 — Live demos, audio, photorealism, and real-time interaction 11:03 — Egocentric first-person view 11:53 — Dialogue with characters in the world 15:10 — Gemini in the loop 17:10 — Fantastical world simulation 19:01 — Genie 4 and robotics 24:11 — What makes Genie 4 shine vs. previous models 25:55 — Scaling and future directions 34:42 — API availability 35:04 — Safety and SynthID in real time 41:36 — Go-to prompts from the team 45:36 — Evaluating world model quality 🔗 Resources: Genie 4 Blog? → https://goo.gle/4ipGTNa Subscribe to Google for Developers → https://goo.gle/developers Speakers: Raia Hadsell, Diego Rivas, Hang Qi, Logan Kilpatrick Products Mentioned: Gemini