All models

Physically grounded 3D world model

Echo-2

Turns text, images, or panoramas into persistent 3D worlds that can be explored and shared in a browser.

Public web app, gallery, and developer APIPublic product
OrganizationSpAItial
CategoryPhysically grounded 3D world model
AvailabilityPublic SpAItial app with free-start messaging, public gallery, and a developer API for asynchronous world-generation jobs.
Last reviewed2026-08-27

01 / Overview

Text and image to 3D world generation, Gaussian-splat delivery, browser exploration, scene editing and sharing.

Turns text, images, or panoramas into persistent 3D worlds that can be explored and shared in a browser.

Inspect geometry at room boundaries, occluded areas, reflective surfaces, and thin objects.

Evaluate the downloadable representation separately from the quality of the hosted viewer.

02 / Strengths

Where it stands out

  • Provides a real consumer creation flow rather than only a research showcase.
  • Outputs a persistent 3D representation that can be viewed from new camera positions.
  • Supports text, image, and panorama inputs plus browser sharing.
  • Offers documented Gaussian-splat outputs through the API.

03 / Boundaries

What not to overclaim

  • Visual 3D consistency does not guarantee production mesh topology, collision, or complete object separation.
  • Physics-based dynamics are described as a future direction, not a current universal capability.
  • Credits, generation time, editing access, and exports depend on the current product plan.

A spatially persistent 3D scene generator; future physics and dynamics should not be claimed as current general simulation.

04 / Access & output

How the model reaches people

Check the live SpAItial app for free credits, paid plans, queue priority, and export allowances.

Access surface
Hosted product with current account plans
SpAItial app
Platforms
Web app · Interactive gallery · Developer API
Outputs
Explorable 3D world · SPZ Gaussian splat · SOG Gaussian splat · Share link
Inputs & compatibility
Text prompt · Single image · Panorama · Web browser

05 / Practical path

How to evaluate it

  1. 01

    Start with a clear text prompt, image, or panorama in the SpAItial app.

  2. 02

    Generate the world, then move away from the input camera to inspect hidden regions.

  3. 03

    Use editing and sharing only after checking what the current account tier includes.

  4. 04

    Confirm SPZ or SOG requirements in the API docs before planning an external viewer workflow.

Primary evidence

Sources