MeshGPT

MeshGPT is a design tool that turns text prompts or images into exportable 3D models. MeshGPT generates textured 3D models from text prompts or images in the browser, with export to GLB, FBX, OBJ, STL and more. It has a free daily tier, commercial use rights and a pay-as-you-go API.

MeshGPT is a browser-based 3D model generator that builds meshes from either a written prompt or an uploaded picture. It targets people who need 3D assets but do not model by hand: indie game developers, anime and VTuber creators, product designers, 3D printing hobbyists and web developers who want richer visuals. The site presents two main entry points, Text to 3D and Image to 3D, plus a separate file converter page for moving models between formats. In practice, a user either describes an object in words or uploads a JPG, PNG or WebP photo, sketch or illustration. The system returns a textured model, with PBR materials generated for text-based results and original colors and fine detail kept for image-based ones. Styles range from realistic to cartoon to low-poly. Generation time depends on the setting: fast models are quoted at roughly 20 to 30 seconds, while the highest quality runs with detailed texture and geometry options take one to three minutes. Results can be previewed on the site before downloading, and retexturing and remeshing tools are mentioned in user feedback. Export coverage is broad. Models can be downloaded as GLB, FBX, OBJ, USDZ, STL, DAE and BLEND files, which covers web and AR, game engines such as Unity and Unreal Engine, Blender, Apple AR and printers. The homepage states that generated models may be used commercially. Developers can also reach generation, AI texturing and remeshing through a REST API billed with pay-as-you-go credits. Among alternatives, MeshGPT sits in the accessible end of the AI 3D market: a free tier with a limited number of daily generations, no credit card requirement and a narrow focus on asset creation rather than a full modeling suite. It suits quick prototyping, game props and printable figures better than precision engineering work, where hand-built geometry and manual control still matter. Teams that need heavy scene editing or rigging will likely pair its output with a conventional 3D application.