# TRELLIS.2 online TRELLIS.2 is an open, MIT-licensed image-to-3D model. This page runs it on hosted GPUs: upload one picture, choose a resolution, and get a GLB with base color, roughness and metallic maps, with nothing to install. > Canonical: https://ai3d.studio/trellis-2 ## Run TRELLIS.2 online ## The model in numbers ### 32, 39 or 45 credits One run at Draft (512), Standard (1024) or High (1536) resolution; Standard is preselected. ### 4 billion parameters The size of the published model, with open weights under the MIT license. ### 1 picture in The model builds a mesh from a single image; it takes no text prompt and no second view. ### 50K to 1M vertices The vertex budget the GLB is reduced to; 200K is preselected so the file still opens quickly in a browser. ## What one run returned A product photo of a toy robot, run at the preselected settings. The figures describe the file the run returned; the viewer loads a compressed web copy. Open the card to orbit the model. ### Toy robot, Standard resolution 199K triangles with a base-color texture and a metal-roughness map; a 7.9 MB GLB after 202 seconds end to end. ## What TRELLIS.2 is The facts below come from the model's published model card, paper and project page. ### Released in December 2025 The code and weights came out in December 2025, with the paper on arXiv the same month and the training code in January 2026. ### Open weights under MIT The model card states the MIT license, which allows commercial use; running it yourself needs an NVIDIA GPU with at least 24 GB of memory, and it is tested on Linux. ### Shape and material in one voxel grid TRELLIS.2 is built on a voxel structure called O-Voxel, which encodes geometry and PBR material (base color, metallic, roughness and alpha) together; models are generated at up to 1536³. ### Good at hard shapes The project page highlights open surfaces, non-manifold geometry and enclosed interior structures, with sharp edges preserved. ### Run here on hosted GPUs AI3D Studio runs the published model through an inference provider and bills credits for the GPU time; it did not develop the model. Each result is saved to your library. ## TRELLIS.2 settings in this generator Every option the form shows, with what it changes. ### Resolution Draft, Standard, High ### Texture size 1K, 2K, 4K ### Vertex count 50K, 200K, 1M ### Clean topology On ### Seed Any number ## Make a model from a picture in three actions The form above is preset to Standard resolution, the middle price. ### Upload one picture A JPEG, PNG or WebP up to 20 MB, with one subject on a plain background and nothing cut off at the edges. ### Pick resolution and texture size Draft, Standard or High sets the price of 32, 39 or 45 credits; the texture size and vertex count are free choices. ### Generate and download The GLB lands in your library when the run finishes; convert it to OBJ, STL, FBX or USDZ in the browser if you need another format. ## How it differs from the first TRELLIS The first TRELLIS, released in December 2024 under MIT, is a different family of models. ### One model instead of several The first release shipped a 1.2B image model and text models from 342M to 2.0B parameters. The new release is one 4B model that takes a picture only. ### Textured meshes instead of splats The first TRELLIS could output Gaussians, radiance fields or meshes. The new model returns a mesh with PBR materials, which is what game engines and slicers open. ### Finer voxel grids It works at 512³, 1024³ or 1536³, the three resolutions this form offers. ## When another model fits better than TRELLIS.2 This model takes one picture and returns a GLB. These cases call for something else. ### [You have several photos](https://ai3d.studio/hunyuan-3d) Hunyuan3D 3.1 Pro reads up to eight views of the same object. ### [You need quads or a T-pose](https://ai3d.studio/rodin-ai) Rodin Gen-2.5 rebuilds the mesh as quads and stands characters in a T-pose or A-pose. ### [You only have a description](https://ai3d.studio/text-to-3d) Text to 3D runs models that start from words instead of an image. ### [You want every option side by side](https://ai3d.studio/compare-ai-3d-models) Credits, inputs, formats and timings for all twelve generation models. ## Known limits of the model Some come from the model itself, some from running it as a service. ### Small holes in raw meshes The model card warns of occasional small holes or breaks in the surface; repair the model before printing it. ### Results vary from run to run The model has not been tuned to human preference, so a different seed or a cleaner picture can change the result a lot. ### GLB is the only native format OBJ, STL, FBX and USDZ are converted from the GLB in your browser for free after the run. ### Slower than the model card The model card lists about 17 seconds at 1024 on an H100 for generation alone. A hosted run adds upload, queueing and download, and the measured Standard run took 202 seconds end to end. ## Questions about TRELLIS.2 Answers about cost, input, speed, licensing and output. ### What is TRELLIS.2? A 4-billion-parameter image-to-3D model released in December 2025 under the MIT license. One picture goes in and a mesh with PBR materials comes out. ### How much does TRELLIS.2 cost here? A run costs 32 credits at Draft, 39 at Standard and 45 at High resolution. Texture size and vertex count do not change the price, and the converter turns the GLB into other formats for free. ### Can TRELLIS.2 make a 3D model from text? No. The model takes a single image only. Generate a picture first, or use text to 3D, which runs models that start from a description. ### How long does a run take? The model card lists about 3, 17 and 60 seconds at 512, 1024 and 1536 on an H100. A hosted run adds upload, queueing and download; the measured Standard run took 202 seconds end to end. ### Is TRELLIS.2 free? The weights are free to download under MIT, but running them yourself needs an NVIDIA GPU with 24 GB of memory and a Linux setup. Here a hosted run costs 32 to 45 credits, needs no setup, and the result is saved to your library. ### TRELLIS.2 or Hunyuan3D: which should I pick? Pick TRELLIS.2 for one picture and an MIT-licensed model at 39 credits. Pick Hunyuan3D 3.1 Pro at 47 credits when you have several views of the object or need up to 1.5M faces and an OBJ copy. ### Who uses TRELLIS.2, and for what? Game and app developers turning concept art or product photos into textured props, and makers who want a model from a picture without setting up a GPU. Characters that must be rigged usually go better through a model with a T-pose option. ### How many credits does TRELLIS.2 use? TRELLIS.2 draws between 25 and 32 credits per run, depending on the model and length chosen in the form. ### What limits should I check before using TRELLIS.2? Generated output from TRELLIS.2 can contain visual, audio, or factual errors. Review each result before publishing, and upload only material you are authorized to use. Where the form accepts a reference image, supported files are JPEG, PNG, WebP up to 20 MB each. ### What happens to the files I upload to TRELLIS.2? An upload is stored against your own account, not pooled with other people's. Deleting the account removes it: the control is in Settings under Security, the request is confirmed by email, and the erasure removes the stored objects it owns. How long anything else is kept, and which processors handle it, is stated in the privacy policy. ### Can I use what TRELLIS.2 makes commercially? Every plan, the free tier included, lets you use what TRELLIS.2 makes commercially: the outputs generated for you are yours. ### How does TRELLIS.2 compare, and what alternatives fit a different workflow? TRELLIS.2: Run TRELLIS.2 online: one picture becomes a GLB with base color, roughness and metal maps, from 32 credits, with no 24 GB GPU or install to set up. Related options are AI 3D model generator: AI 3D model generator with 12 named models: start from one image, a prompt, up to eight views or a sketch. From 20 credits; GLB, FBX, OBJ, STL or USDZ out, Rodin AI: Run Rodin AI, Hyper3D's Rodin Gen-2.5, in your browser: a picture to a GLB with PBR maps, quad or triangle meshes and a T-pose option, at 50 credits a run, Image to 3D: Image to 3D in AI3D Studio: upload one photo or render and get a textured GLB from Tripo, Hunyuan3D, TRELLIS.2, Rodin, Meshy, Hitem3D or Seed3D, Hunyuan 3D: Run Hunyuan 3D online: Tencent's Hunyuan3D 3.1 Pro turns a picture into a GLB and OBJ with up to 1.5M faces for 47 credits, and 3.1 Rapid drafts one for 29, Text to 3D: Text to 3D with Tripo, Hunyuan3D, Rodin and Meshy: describe an object in a prompt and get a textured 3D model from 26 credits, with the model named, and Multi-view to 3D: Multi-view to 3D in AI3D Studio: upload front, side and back views and get a model whose hidden sides are measured, not guessed. Up to 8 views. Choose by the required input, delivered output, controls, limits, current availability, and credit cost; the related-tools section links each canonical page. ## Turn a picture into a model with TRELLIS.2 Upload an image above and run it at Standard resolution. ## Related tools ### [AI 3D model generator](https://ai3d.studio/ai-3d-model-generator) ### [Rodin AI](https://ai3d.studio/rodin-ai) ### [Image to 3D](https://ai3d.studio/image-to-3d) ### [Hunyuan 3D](https://ai3d.studio/hunyuan-3d) ### [Text to 3D](https://ai3d.studio/text-to-3d) ### [Multi-view to 3D](https://ai3d.studio/multiview-to-3d) _Last updated October 8, 2026_