Hi3D launches Hi3D Agent, an all-in-one AI agent for 3D creation
Hi3D launched Hi3D Agent on September 21, combining a large language model with the company's 3D tools for geometry, textures, mesh work, and model refinement. The agent includes reusable skills for tasks like optimization, animation, and converting high-poly models into playable demos, and is positioned to carry assets through game, animation, design, and 3D-printing workflows. A launch pairs the product with a creator rewards initiative.
- Hi3D launched Hi3D Agent on September 21, pairing a large language model with its 3D tools for geometry, textures, mesh work, and model refinement.
- The agent includes reusable skills for optimization, animation, and converting high-poly models into playable demos, per the company.
- A Windows client is listed on the product page, with a workflow for generating, revising, and finishing models and carrying assets into game, animation, design, and 3D-printing steps.
Entities
Related
Mixar: AI agent for 3D scene creation in Blender
Mixar is a tool that integrates AI agents with Blender to automate 3D scene creation and editing. It uses LLMs for tool use and spatial understanding, addressing challenges like scene graph access, parallelism, and verification. It aims to be a 'Cursor for 3D', providing a single access point for generative inference and handling tasks like UV mapping and export.
Mixar is a tool that integrates AI agents with Blender to automate 3D scene creation and editing. It uses LLMs for tool use and spatial understanding, addressing challenges like scene graph access, parallelism, and verification. It aims to be a 'Cursor for 3D', providing a single access point for generative inference and handling tasks like UV mapping and export.
GPT-6 Astra + 3DAIStudio MCP and Tripo P2 build Rocket League-style 3D scene via agentic pipeline
A demonstration connects GPT-6 Astra to the 3DAIStudio MCP and Tripo AI, letting the model autonomously generate 3D assets and assemble a playable scene. Astra generates reference images for assets it decides it needs, converts them to game-ready 3D models via Tripo P2, and imports them back while building the scene. The example produced a Rocket League-style environment, showing agent-driven 3D content pipelines rather than manual artist workflows.
A demonstration connects GPT-6 Astra to the 3DAIStudio MCP and Tripo AI, letting the model autonomously generate 3D assets and assemble a playable scene. Astra generates reference images for assets it decides it needs, converts them to game-ready 3D models via Tripo P2, and imports them back while building the scene. The example produced a Rocket League-style environment, showing agent-driven 3D content pipelines rather than manual artist workflows.
AI agent for 3D printing in OrcaSlicer
LayerPilot is an AI agent integrated into OrcaSlicer that automates 3D printing preparation. Users can give natural language commands like 'prepare this model for printing', and the agent orients the model, adjusts settings, generates supports, splits large models, and slices. It aims to streamline the workflow from prompt to print-ready project, with a planned Meshy integration for generating 3D models from text.
LayerPilot is an AI agent integrated into OrcaSlicer that automates 3D printing preparation. Users can give natural language commands like 'prepare this model for printing', and the agent orients the model, adjusts settings, generates supports, splits large models, and slices. It aims to streamline the workflow from prompt to print-ready project, with a planned Meshy integration for generating 3D models from text.
AI-driven engine that generates 3D game assets from a reference image via iterative LLM workflow
An engine that structures an LLM into a workflow of iterating and judging its own output to create 3D assets for games. The user supplies a reference image and a prompt, and Claude Opus 5 works through the engine for 3-4 hours per asset. It targets game developers who want to generate assets without manual modeling.
An engine that structures an LLM into a workflow of iterating and judging its own output to create 3D assets for games. The user supplies a reference image and a prompt, and Claude Opus 5 works through the engine for 3-4 hours per asset. It targets game developers who want to generate assets without manual modeling.