Breaking

Higgsfield Uses GPT-6 Astra to Rebuild Babylon’s Lost Gardens in 3D

September 5, 2026 at 01:03 EDT

Foundation Models
Media Generation
Software Dev & Coding

Higgsfield has presented a digital 3D reconstruction of the Hanging Gardens of Babylon, combining GPT-6 Astra for geometry with Seedance 2.5 for the final video pass on Higgsfield Supercomputer. The demonstration is part of a broader attempt to visualize the Seven Wonders of the Ancient World from descriptions rather than surviving structures.

SEP 5, 2026 · HIGGSFIELD

A Lost Wonder, Rebuilt in 3D

GPT-6 Astra builds the geometry of the Hanging Gardens of Babylon, while Seedance 2.5 produces the final visual pass on Higgsfield Supercomputer.

The scale of the story

ancient wonders

2,500

years as a description

featured reconstruction

Astra benchmark context

FrontierMath Tier 4 published scores

97.6%

GPT-6 Astra

83.0%

GPT-5.6 Sol

Astra leads the published comparison, but this is not a score for the Babylon reconstruction.

From description to final pass

Description

A textual account of a lost site.

→

GPT-6 Astra

Builds the scene geometry.

→

Seedance 2.5

Generates the final pass.

What the demo shows

A description-led pipeline can turn an uncertain historical subject into a coherent visual interpretation.

What remains unproven

The scene has no published task score or archaeological validation, so its historical fidelity remains interpretive.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.