Krux
February 10, 2026
DreamTech's Neural4D-2.5 Slashes 3D Content Costs by 95%
Published: February 10, 2026 at 12:45 AM
Updated: February 10, 2026 at 12:45 AM
100-word summary
DreamTech just dropped Neural4D-2.5, a foundational AI model that turns text and images into production-ready 3D content—at 1 yuan per model. Unveiled in Shenzhen last December and officially announced February 8, it uses a "Native 3D Attribute Grid" architecture with TEXTRIX tech to generate high-fidelity textures, geometry, and watertight models ready for 3D printing. The kicker? Training costs run 1/30th of industry norms, inference at 1/20th, slashing overall production costs by roughly 95%. This could democratize 3D content creation for gaming, manufacturing, and metaverse applications at unprecedented scale.
What happened
DreamTech just dropped Neural4D-2.5, a foundational AI model that turns text and images into production-ready 3D content—at 1 yuan per model. Unveiled in Shenzhen last December and officially announced February 8, it uses a "Native 3D Attribute Grid" architecture with TEXTRIX tech to generate high-fidelity textures, geometry, and watertight models ready for 3D printing. The kicker? Training costs run 1/30th of industry norms, inference at 1/20th, slashing overall production costs by roughly 95%.
Why it matters
This could democratize 3D content creation for gaming, manufacturing, and metaverse applications at unprecedented scale.