
Agora-1
Odyssey’s multi-agent world-model preview lets multiple human or AI participants share and act inside the same generated simulation in real time.
An interactive world model designed to stream generated video as people act inside the scene.
Explore platform ↗Visko Orbis
A world stream has to preserve enough state for the next input and meet a playback deadline for each output chunk. That makes the recovery behavior as important as the initial visual quality.
The interesting question for a viewer is whether a change of direction still fits the same world after a pause. This is a stronger test than a single smooth pan. If a landmark changes when it goes out of frame, a shared audience can lose track of the story even when playback never stops.
This is a directory note, not a promise that the third-party channel is always live. Availability, moderation, and content are controlled by the linked project.
Start with the official Orbis entry point. If you can open a session, run a short sequence with a pause, a direction change and a return to an earlier view. Record whether motion responds, whether the world stays legible and what happens if the session stalls.
Best for: Creators and developers studying interactive scenes, continuity and live world model interfaces.
Access note: Use Visko’s own site for the current demo and access terms. Hosted session availability and performance depend on the actual service and were not checked here.
2026-11-20: Official model page and a developer deployment write-up reviewed. We did not start a session or verify latency and access.

Odyssey’s multi-agent world-model preview lets multiple human or AI participants share and act inside the same generated simulation in real time.

A gallery of R2-powered interactive worlds and character experiences where prompts and controls can influence a continuing audiovisual session.

Sequencer’s directory of continuous AI video worlds. Viewers vote, chat, suggest prompts, and help direct a shared timeline.