Join Genie 4 Waitlist: Project Genie Access Guide

Google DeepMind released Genie 4 on October 27, 2026. The new world model generates photorealistic interactive environments with native audio, real-time steering, and better physics. It builds on Gemini Omni for visual grounding and other Gemini models for reasoning and continuity.

Access remains limited. Trusted testers have it now. Google AI Ultra subscribers get it next in Project Genie, followed by early access partners on the Gemini API and AI Studio. You can join the official waitlist today.

This guide covers the exact steps to join, what to expect, and how to write prompts that work once access arrives. All details come from the DeepMind announcement and the Project Genie prompt guide.

What Genie 4 Adds

Genie 4 improves four areas over earlier versions.

It produces more realistic visuals by anchoring simulations in Gemini Omni. Physics handling covers aerodynamics, collisions, and natural human movement. Users steer scenes live with natural language prompts while the simulation runs. Native audio generates ambient sounds, music, and interactive dialogue in sync with the visuals.

Gemini acts as the director. It adapts prompts to the current scene and maintains coherence between inputs. Sessions can stay consistent longer than previous models.

The official demo video shows real-time keyboard navigation plus text prompts that change weather, add characters, or alter the camera.

Join the Genie 4 Waitlist

DeepMind directs interested users to a Google Form. The link appears on the announcement post and the DeepMind blog.

1. Open the official form at forms.gle/WQnSNEM3omF7Uruu7.
2. Provide the requested details. Expect questions about your use case, such as creative tools, robotics research, or agent training.
3. Submit the form. Confirmation is usually immediate via email.
4. Watch your inbox for updates. Rollout happens in phases after trusted-tester feedback.

Project Genie currently runs on Genie 3 for Google AI Ultra subscribers in supported regions. Genie 4 arrives later for the same group. Eligibility typically requires an active Ultra subscription and age 18 or older.

If you already use Project Genie, the same account should receive Genie 4 once the phased release begins. No separate signup is needed beyond the waitlist for priority.

Prepare Effective Prompts

Project Genie accepts text prompts and optional reference images. Genie 4 inherits the same interface while adding stronger real-time control.

Start with a clear environment description. Name the setting, lighting, weather, and key objects. Add a character if you want navigation from a first-person or third-person view. Specify movement style, such as walking, flying, or driving.

For real-time steering, type short instructions while the world runs. Examples include “change the weather to rain,” “add a second person on the right,” or “move the camera higher.” Gemini rewrites the prompt to fit the existing scene.

Audio instructions work too. Request specific sounds, such as “wind through trees and distant waves,” or dialogue between characters. The model generates matching audio on the fly.

The earlier Project Genie tip article lists four practical rules:

• Describe the environment in detail before adding characters.
• Define the character’s appearance and movement.
• Use reference images when you need a specific starting frame.
• Iterate with short steering prompts rather than rewriting the whole scene.

These rules transfer directly to Genie 4. Test them on Genie 3 first if you have Ultra access. The muscle memory helps once Genie 4 arrives.

What Happens After Access

Once you receive Genie 4 in Project Genie, open the web app and start a new world. Upload an image or type a prompt. Navigate with keyboard controls (W, A, S, D for movement; mouse or arrow keys for camera). Type additional prompts in the text field to steer.

Early access partners on the Gemini API and AI Studio will call the model through standard endpoints. Pricing and exact model IDs have not been published yet. Check AI Studio for availability notes.

For creative work, combine Genie 4 outputs with other Gemini tools. Generate still frames with Nano Banana, then expand them into interactive worlds. Or feed simulation screenshots into Gemini for analysis or storyboarding.

If you build agents or robotics systems, the safe simulation environments matter more than consumer play. Genie 4 is positioned for those training loops as well.

Tips for Better Results

Keep initial prompts under a few sentences. Long, dense instructions can reduce coherence. Add detail through follow-up steering prompts.

Test audio by enabling sound in the player. Note which ambient sounds and dialogue styles the model handles cleanly.

Save promising starting frames as reference images for later sessions. Consistency improves when you reuse the same visual basis.

If a scene drifts, reset and try a tighter environment description. The model performs best when the core layout is clear from the start.

For developers planning API use, review the Gemini API documentation for multimodal endpoints. Genie 4 will appear as an additional model once partner access opens.

Conclusion

Genie 4 advances interactive world simulation with native audio and real-time steering. Access starts with trusted testers and expands to Google AI Ultra subscribers in Project Genie, then to API partners.

Join the waitlist now via the official form. Practice prompt writing on existing Project Genie access if you have it. When Genie 4 arrives, the same techniques produce more realistic and controllable results.

For related Gemini workflows, see our guide on building full-stack apps with Gemini 4 Argon in AI Studio at /blog/vibe-code-full-stack-apps-with-gemini-4-argon/.

Sources

• Google DeepMind: Genie 4 announcement (October 27, 2026)
• blog.google: Introducing Genie 4
• Google: How to write prompts for Project Genie
• Official YouTube demo from Google DeepMind