Hi3D has released Hi3D V1.0, formerly Sparc3D, as an open-source model. Source code and weights can now be downloaded by anyone, and the official site lets users generate 3D models for free.

Psst!Add AlexTech.ai asPreferred Source on Google

The model had already drawn attention in 2025 for its high-resolution geometry reconstruction. For this version, the company expanded the training dataset from about 500,000 to two million curated 3D assets and tested it in real production workflows. According to the company's reported results, progressive geometry reconstruction reaches an effective sparse-grid resolution of up to 1536³, with better preservation of fine structures and geometric consistency across viewpoints. Hi3D also reports a 90% slicing success rate, which concerns preparing files for 3D printing.

Hi3D V1.0 official product photo

Image: hi3d-official — https://github.com/hi3d-official/Hi3Dv1.0/tree/main

The open release aims to give a shared technical base to sectors such as 3D printing, product and e-commerce, industrial design, VR/AR/XR, animation and film. Companies and developers can deploy the model privately, integrate it into existing workflows and build specialised products under the applicable license.

The move fits a broader strategy: Hi3D continues to develop proprietary models, with Hi3D V3.0 presented as the first commercially available AI 3D model with 2048³ voxel resolution. Alongside the open model, the company introduced Hi3D Agent, a creator-first agent that pairs a large language model with its proprietary 3D model, and an MCP integration that exposes its capabilities to external agents. The release also comes with a reduction of over 50% in credit consumption.

Hi3D V1.0 official product photo

Image: hi3d-official — https://github.com/hi3d-official/Hi3Dv1.0/tree/main

The case reflects a trend AlexTech has been following for months: open-weight models competing with closed ones. In 3D, long dominated by proprietary tools, a downloadable model at this level lowers the barrier for anyone who wants to experiment or build their own services.