Meshy

What is Meshy?

Meshy is an AI-powered 3D asset generation platform designed for game development, 3D design, AR/VR, and 3D printing. Users can create models from text, a single image or multiple images, or dialogue prompts, and then proceed with tasks such as topology creation, texture application, rigging, animation, and export.

The official workspace is currently divided into six modules: Image, 3D Model, 3D Printing, Animate, Scene, and Video. It also offers 3D Agent, plugins, REST API, API Playground, and MCP, covering the entire process from idea generation to production.

Convert text to 3D text to create 3D models

After the user describes the object’s shape, style, material, proportions, and purpose, Meshy first generates a geometric preview, followed by a detailed version with textures. The API also follows a Preview and Refine approach, allowing developers to assess the shape first before investing resources in adding textures.

The prompt should focus on a single subject and specify the perspective and purpose. Excessively complex scenes can cause multiple objects to overlap with one another; precise mechanical structures and accurate dimensions still require traditional modeling techniques.

Convert image to 3D image to generate 3D.

Image to 3D converts product images, character concept art, or photos into 3D models. Reference images that are square in shape, have a clean background, a well-defined subject, and even lighting are generally more suitable for this conversion process.

A single image cannot provide information about the back side or areas that are obscured; therefore, AI has to estimate those invisible areas. Multiple views can reduce uncertainty, but they do not guarantee engineering precision.

Preliminary concept diagram for Text to Image

In the Image workspace, it is possible to use models such as Nano Banana Pro to generate concept diagrams from text; these diagrams can then be modified further using text, before the final image is sent to Image to 3D. Users can create 1 to 8 variants at a time, and more images will require more Credits.

Meshy 3D Agent

3D Agent offers a conversational workflow that enables users to discuss ideas, create models, adjust directions, and handle the final processing steps. It reduces the learning curve associated with using its functions, but the resulting assets still need to be viewed in a 3D viewer and checked in the target software.

Smart Topology Intelligent topology

Smart Topology is a fast topology model designed for gaming and real-time rendering; according to the official documentation, it can generate a topology in about 10 seconds and emphasizes grid optimization. It is suitable for creating prototypes and processing large numbers of assets, but manual topology work may still be required for hero characters and areas that undergo deformation.

Meshy 6 and Meshy 5

Meshy 6 is aimed at delivering final assets of higher quality, with an official reference time of around 2 minutes; Meshy 5 is better suited for rapid concept iteration, taking about 45 seconds.

For the API, you can choose a specific version or the latest one; the latest version currently refers to Meshy 6.

Upgrading the model version may alter the performance of the prompts as well as the topology; therefore, the production process should use a fixed version while retaining test samples.

Remeshing and polygon control

Remesh can convert the generated mesh into a topology based on triangles or quadrilateral faces, and it allows setting a target number of faces. The Text to 3D API currently supports around 100 to 300,000 target polygons; the actual number may vary depending on the complexity of the geometry.

Reducing the number of polygons results in the loss of outlines and fine details. Game and web assets should have LOD levels determined based on distance from the viewer, rather than focusing solely on minimizing the number of polygons.

AI Texture generation

Users can create models for Meshy, or generate textures for their own FBX, OBJ, GLB, USDZ, STL, BLEND, and 3MF models. The system supports text, reference images, and preset styles, and it handles projection and UV directions automatically.

PBR material

Textures can generate PBR maps such as Albedo, Roughness, Metallic, Normal, and Emission; they are suitable for use in Unity, Unreal, and Blender. The website offers texture options in 2K, 4K, and 8K resolutions – the higher the resolution, the larger the file size.

Automatic UVs may not be ideal in terms of seams and density. Products and characters in close-up views require checking of UVs, normals, and material channels using DCC software.

Automatic binding

AI Auto Rigging can identify Humanoid, Quadruped, or Smart Rig types and generate bone hierarchies as well as skinning weights. A clear T Pose or A Pose usually helps to improve accuracy.

Automatic binding is suitable for prototypes and regular characters; for special body structures, costumes, and facial deformations, manual bone adjustment and weighting are still required.

Over 600 animations

Once the binding is completed, more than 600 preset animations for walking, attacking, waiting, etc., can be applied, and FBX or GLB files containing the skeleton structure can be exported. The naming convention of the skeletons is compatible with Mixamo’s standards, which facilitates further manipulation of them.

Bone Editor

The built-in editor enables the adjustment of bone and skinning weights, after which the model can be imported into Blender, Maya, Unity, Unreal, Godot, or other tools. After exporting, it is necessary to test the joint limits, foot sliding, and mesh penetration.

3D printing

Meshy offers a workflow that goes from generating the model to creating the print file; STL is suitable for monochrome geometries, while 3MF can store color and multi-material information. Before printing, it is still necessary to check for completeness, wall thickness, overhangs, supports, and actual dimensions.

Scene and Video

The Scene module is used to combine multiple 3D assets, while the Video module helps in creating content for display. They are suitable for quick previews and distribution; however, for complex scene layouts, lighting effects, and video post-production, professional tools are still necessary.

Comparison of export formats

Blender, Unity, and other plugins

The official documentation provides information on plugins for Blender, Unity, and other tools, allowing users to utilize the creation and import processes within their familiar DCC applications or engines. When installing such plugins, it is necessary to download them from the official documentation; third-party plugins with the same name available on GitHub may not necessarily be maintained by Meshy.

Meshy REST API

The API covers Text to 3D, Image to 3D, Texturing, Remeshing, and Animation. After submitting an asynchronous task, developers can check its status; upon success, they receive the addresses of the model, textures, and thumbnails.

Failed tasks usually have their corresponding Credits refunded.

The API key must be stored on the server, and the download link may have an expiration date. The application should keep track of the task ID as well as its status – Pending, In Progress, Succeeded, Failed, or Canceled – and save this information before the file expires.

API Playground and MCP

API Playground enables testing of requests without the need to develop a complete application. The official MCP Server allows AI agents to utilize 3D generation capabilities, making it suitable for integrating Meshy into automated creation and game asset workflows.

Before generating batches via the agent, it is necessary to limit the Credits, the number of tasks, and the available functions, in order to prevent errors that could lead to excessive consumption.

Comparison of Meshy plans and prices

The amount displayed on the page is affected by the currency, whether payment is monthly or annual, and by any ongoing promotions. The monthly Credits are reset each month and do not accumulate;

Permanent Credits purchased additionally or as rewards do not expire.

Licensing of free and paid assets

Assets generated under the free plan are licensed under CC BY 4.0, allowing them to be used and sold as long as proper attribution is given; paid premium subscribers own the assets they generate and can distribute and sell them.

Regardless of the package chosen, the images, characters, and brands entered by the user must have legitimate rights.

Education discount

Current students and educators can apply for a discount on the Meshy Education Plan. Eligibility, location, amount of the discount, and renewal terms must be verified on the education program page.

Refunds and cancellations

Subscriptions can be canceled or upgraded in the settings. Due to the cost of GPUs, refunds are generally not provided by the vendor; exceptions are considered only under specific circumstances, with a processing time of approximately three working days.

Data and Privacy

The official website states that the data of web applications is stored in AWS in the United States, and it is not shared or used for training without prior consent. When dealing with customer models, confidential roles, and trade secrets, companies should still review the contracts, determine the retention periods, and establish procedures for deletion.

GitHub and the open-source status

Meshy’s generation model, Web workspace, API backend, and Credits system are not open-source products. Projects such as meshy-ai and Unity-AI-Meshy listed in the real-time GitHub results have not been verified through the official website documentation as being the official source code for the core platform.

Some plugins or community connectors may have their code made public, but this does not allow Meshy to be classified as open source or suitable for private deployment.

Meshy usage tutorial

Complete a basic task.

1. Clarify the purpose of the image, its dimensions, style, and requirements regarding commercial licensing.
2. Upload the original images for which you have usage rights to Meshy, or provide clear instructions;
3. Select Text to 3D to create a 3D model, generating low-cost candidates first;
4. Use Image to 3D to generate 3D versions with adjusted composition, subject, and details;
5. Zoom in to examine faces, hands, text, edges, logos, and artifacts;
6. Export after confirming the color, resolution, and licensing, while retaining the original image;

Create reusable professional workflows

1. Create a list of brand colors, fonts, layout elements, and elements that are prohibited.
2. Separate tests were conducted for Text to 3D for generating 3D models, Image to 3D for creating 3D models from images, and Text to Image for producing preliminary concept diagrams.
3. Use the same set of representative samples to compare quality, speed, and cost;
4. Complex edges, text, and images of key products should be handed over for manual refinement.
5. Standardize naming, dimensions, and review status;
6. Batch processing and release are carried out after random inspections;

Which users is it suitable for?

• Developers who can create game characters, items, and environmental assets quickly;
• Designers who convert concept diagrams into 3D models;
• 3D artists who need PBR textures, binding, and animation;
• Teams that create AR, VR, and web-based 3D content;
• Makers who create printable models and customized gifts;
• Developers who generate assets in bulk through APIs and MCP.

Product advantages

• Three ways to create 3D content: text, images, and dialogue;
• An integrated process ranging from grids and textures to animated bindings;
• It supports the mainstream formats for games, web pages, AR, and 3D printing;
• Provides plugins, REST APIs, a Playground, and MCP;
• The free version provides 100 Credits per month;
• The paid plan provides clearer ownership of the generated assets.

Restrictions and Precautions

• Generation models may have issues with topology, backside, UVs, and animation deformation, so they cannot be guaranteed to be ready for production.
• High-frequency generation consumes a large amount of Credits, and the monthly Credits cannot be carried over.
• Free assets must be credited, and the user is responsible for any copyright and portrait rights related to those materials.
• Professional games, as well as products manufactured and printed, should be thoroughly checked in the target software before use;

Frequently Asked Questions

Is Meshy free?

There is a free option: 100 Credits per month, no credit card required; the output is under CC BY 4.0.

How much is Meshy Pro?

The current reference price is $20 per month or $240 per year, with 1,000 Credits per month.

Can Meshy automatically bind and add animations?

Yes, it supports Humanoid, Quadruped, and Smart Rig, and offers over 600 animation presets.

Does Meshy provide an API?

It offers REST APIs, a Playground, and MCP, covering generation, textures, remeshing, and animation.

Is Meshy open source?

The core platform is not open source; no GitHub repository containing the complete source code for the official models and Web services has been found.