AI StreamRADAR
Submit a Project
Menu

AlayaWorld

An interactive world-model demo on Hugging Face Spaces: a starting image, a scene prompt and a camera direction become a video world you keep steering.

Last checked (just now)Method: Editor verifiedConfidence: medium
Try ItVisit Website

Spotted an error or a status that looks wrong? Report an issue.

A / Editor’s summary

What AlayaWorld is

AlayaWorld is the name on the Hugging Face Space https://huggingface.co/spaces/AlayaLab/alayaworld-demo, published by the AlayaLab organization. The Space was created on 15 September 2026, and its public Gradio app was running when checked on 20 October 2026, with the Hugging Face API reporting a RUNNING state on ZeroGPU hardware (zero-a10g). Anyone can open the demo; there is no broadcast to watch.

The demo is interactive: it takes a starting image, a scene prompt and a camera direction, then generates a video world while you keep steering, with spatial memory presented as its headline feature. The code is public at https://github.com/AlayaLab/AlayaWorld. The Space README names the model as AlayaLab/AlayaWorld; no separate checkpoint page is tracked here, and no frame-rate or latency figure is verified.

AI Live Index, a third-party directory, lists the demo as a public beta on 9 October 2026. This listing is on demand — an interactive session you start yourself, not a live channel — in the same category as MIRA (/platforms/mira/) and WorldPlay2 (/platforms/worldplay2/).

Interactive Worlds
B / How it works

How the experience flows

  1. Open https://huggingface.co/spaces/AlayaLab/alayaworld-demo and wait for the Gradio app to load.
  2. Provide a starting image, a scene prompt and a camera direction, then keep steering as the demo generates the video world around those inputs.
  3. Treat a session as an on-demand generation: there is no shared broadcast, and performance figures are unverified.
C / Status & experience

Status, as checked

AlayaWorld demo Space
Watch
https://huggingface.co/spaces/AlayaLab/alayaworld-demo
Last checked
Valid until
Method
Editor verified
Confidence
medium
Website
Online (HTTP 200)

Site probe at 2026-11-01T17:43:01.231Z: HTTP 200 (online). Recorded as on demand — Interactive on-demand world-model demo on Hugging Face Spaces; there is no broadcast endpoint. Strategies empty, monitor disabled, editor status on demand — never a live broadcast.

Available to watch on demand; not a live broadcast.

Website reachability and broadcast status are recorded separately; an HTTP 200 never counts as live. Expired checks degrade to Stale, detector failures to Unverified. Methodology.

D / Why it matters

Why we track it

Places a publicly playable world-model demo with public code among the interactive-world listings, recorded as on demand so it is never confused with the verified live channels.
E / Models & technology

Models behind the output

Not publicly disclosed.

F / Updates timeline

Timeline

No dated updates recorded yet.

G / Sources & verification

Where each fact comes from

Editorial review: . Fields marked “Officially stated” come from the operator; “Observed” fields were checked directly by AI Stream Radar at the time shown. Quoted text is marked and attributed to its source.

  1. AlayaWorld demo — Hugging Face Space
    Observed by AI Stream RadarAlayaLabObserved Confidence: high
    “RUNNING”

    HTTP 200 on 20 October 2026. The Hugging Face API reported the Space as RUNNING on ZeroGPU hardware (zero-a10g). The Space was created on 15 September 2026; that creation date is recorded as a Space fact, not as a launch date, so launch_date stays null with unknown precision. The public Gradio app takes a starting image, a scene prompt and a camera direction and generates a video world while the visitor keeps steering. Anyone can open the demo; there is no broadcast endpoint, so the listing is on demand. No frame-rate or latency figure is verified, and ZeroGPU queueing behaviour is unverified.

    namewebsite_urlorganization_idplatform_kindmedia_typeinteraction_modesaccess_typestreams.main.watch_url
  2. AlayaWorld on GitHub
    Officially statedAlayaLabObserved Confidence: high

    The public repository is at https://github.com/AlayaLab/AlayaWorld. The Space README names the model as AlayaLab/AlayaWorld; no separate checkpoint page is tracked here. The GitHub API reports the repository license as Other (SPDX NOASSERTION) and the Hugging Face Space README license as other, so no confirmable open-source license exists and is_open_source is recorded as unknown; public code is not by itself an open-source statement. The demo is an interactive session, not a broadcast.

    github_url
  3. AI Live Index — AlayaWorld
    Reported by third partyAI Live IndexObserved Confidence: medium
    “PUBLIC BETA”

    AI Live Index, a third-party directory, lists the demo as a public beta on 9 October 2026 and describes a starting image, a scene prompt and a camera direction producing a video world with spatial memory as the headline feature. A directory listing is not a measurement.

    maturity
H / Related platforms

Agora-2

Odyssey's playable world-model research preview lets up to 20 people or agents interact in one generated world.

Research PreviewMulti-player Shared Input
Verified
AIXplorer — official preview

AIXplorer

A starting image becomes a world you steer by typing or speaking — on demand, not a live broadcast.

Interactive WorldH3 MaxFree-form PromptVoice Input
Verified

ALIBI

A scene-driven manor mystery (The Last Light) you enter on demand — not a continuous AI livestream.

Interactive WorldMultiple Choice
Verified