Hunyuan 3D
Hunyuan 3D is a family of open weight generative artificial intelligence models from Tencent that turn text prompts, single images, sketches, and other inputs into ready to use three dimensional assets, complete with geometry and physically based textures, in minutes rather than the days a human modeller might need.[1] It belongs to the broader Hunyuan suite of foundation models, which also includes large language models, the HunyuanVideo text to video system, and the HunyuanImage diffusion model. The lineage began in November 2024 with the open release of Hunyuan3D 1.0 and progressed through Hunyuan3D 2.0, 2.1, 2.5, and 3.0 over the course of 2025.[5][3] On 25 November 2025 Tencent announced the global launch of the Hunyuan 3D creation engine and an English language API on Tencent Cloud, bringing the hosted creation tools to creators worldwide.[1][2]
At the time of launch the project had crossed three million cumulative downloads on Hugging Face across its open weight releases, which Tencent says makes it "one of the world's most popular 3D open-source models."[1][8] Tencent claims that more than 150 enterprises in Mainland China had integrated the technology before the global rollout, with Unity China, the 3D printer manufacturer Bambu Lab, and the AI content platform Liblib among the named partners.[1][15] The company frames the value proposition bluntly, saying the engine "cuts production time from days or weeks to just minutes."[1] Hunyuan 3D 3.0, the production model at the global launch (offered through Tencent's platform and API rather than as open weights), supports geometric resolutions of 1,536 cubed voxels and produces meshes intended for direct use in game engines, additive manufacturing, virtual reality, and e commerce visualization.[16][17]
What is Hunyuan 3D used for?
Tencent began publishing generative AI research under the Hunyuan brand in 2023. By late 2024 the company had shipped a billion parameter class text to image model, a Mixture of Experts language model, and the HunyuanVideo system for short clip generation. The 3D track grew out of internal needs at Tencent Games, where, in the words of the Hunyuan3D Studio paper, the creation of high-quality 3D assets "has long been characterized by labor-intensive and specialized workflows."[12] The Hunyuan3D project set out to compress that cycle to seconds while still emitting meshes and textures that fit existing digital content creation pipelines.[23] In practice the models are used for game asset prototyping, 3D printing, virtual and augmented reality scenes, product visualization for e commerce, and rapid concept iteration.
The project entered a crowded field. Stability AI had released Stable Zero123 in December 2023 and, with Tripo AI, TripoSR in early 2024, the startup Tripo AI had shipped commercial text to 3D systems, Meshy had built a subscription product around image to 3D, and the academic Trellis system from Microsoft Research had demonstrated structured latent diffusion for shape and texture. Tencent's contribution was scale: large diffusion transformers, a paired texture model, and an open licence that permitted commercial use across most of the world.[9]
When was each Hunyuan 3D version released?
The table below lists the publicly disclosed Hunyuan 3D releases through May 2026. Dates refer to the first public availability of either weights, a technical report, or a hosted API.
| Version | Released | Key change |
|---|---|---|
| Hunyuan3D 1.0 | November 2024 | Two stage framework, multi view diffusion plus feed forward reconstruction, text and image input |
| Hunyuan3D 2.0 | 21 January 2025 | Hunyuan3D DiT for shape and Hunyuan3D Paint for texture; first community licence release |
| DiT v2 0 Fast | 3 February 2025 | Distilled guidance variant for faster sampling |
| Hunyuan3D 2.0 Turbo | 19 March 2025 | Step distilled checkpoints for production latency |
| Hunyuan3D 2.5 | 23 April 2025 | 10 billion parameter shape model, 1,024 cubed geometric resolution, 4K PBR textures |
| Hunyuan3D 2.1 | 13 June 2025 | First fully open source PBR pipeline, training code and full weights released |
| Hunyuan3D PolyGen | 7 July 2025 | Autoregressive mesh model with clean quad and triangle topology |
| HunyuanWorld 1.0 | July 2025 | Open source 3D world generation from text or image |
| Hunyuan3D Omni | September 2025 | ControlNet style multi condition input including pose, point cloud, voxel, and bounding box |
| Hunyuan3D 3.0 | 16 September 2025 | 1,536 cubed voxel resolution, 3.6 billion voxels, rough to fine 3D DiT |
| Hunyuan 3D Studio | September 2025 | Invitation only end to end pipeline with UV unwrapping, rigging, and skinning |
| Hunyuan 3D creation engine global launch | 25 November 2025 | Public web platform plus English Tencent Cloud API |
| HunyuanWorld 1.5 (WorldPlay) | December 2025 | Real time world generation for interactive playback |
| Hunyuan 3D 3.1 | January 2026 | API model option adding top, bottom and 45 degree view inputs; API only[32] |
| HunyuanWorld 2.0 | April 2026 | Updated 3D world model |
How does Hunyuan 3D work?
The core Hunyuan3D pipeline separates geometry from appearance. A shape model produces an untextured mesh, and a texture model paints that mesh with view consistent material maps.[3] This decoupling is similar to the design used by Trellis, CRM, and other contemporary systems, but Tencent scaled both stages to large diffusion transformer backbones. The 2.1 technical report describes the split directly, noting that "Hunyuan3D 2.1 separates shape and texture generation into distinct stages" as "a more advanced strategy proven effective upon previous large reconstruction models."[31]
Shape generation
The shape model is named Hunyuan3D DiT.[3] It is a flow based diffusion transformer that conditions on a reference image or a text embedding and denoises an implicit volumetric representation. The 1.0 release used a multi view diffusion stage followed by a feed forward sparse reconstruction network, but starting with 2.0 the project moved to a single stage native 3D diffusion.[5][3] Across versions Tencent shipped several sizes: a 0.6 billion parameter mini variant, a 1.1 billion parameter standard variant in the 2.0 family, and a 3.3 billion parameter variant for 2.1.[8][4] The 2.5 release scaled the shape model to roughly 10 billion parameters and lifted the output resolution to 1,024 cubed voxels.[18] Hunyuan 3D 3.0 reaches 1,536 cubed, which Tencent describes as 3.6 billion voxels, and uses a coarse to fine sampling schedule that first blocks out gross form and then refines micro detail.[16][17]
Texture and material synthesis
The texture model is named Hunyuan3D Paint. It takes the mesh from the shape stage and renders consistent textures across multiple viewpoints, then bakes those views into UV space.[3] Versions 1.0 and 2.0 produced albedo only RGB textures. Version 2.1 introduced physically based rendering outputs, generating albedo, metallic, and roughness maps.[4][31] The 2.1 paint model reports 2 billion parameters, while the shape model reports 3.3 billion.[4] A separate Hunyuan3D Delight model handles relighting and shading correction.
Mesh topology
The original Hunyuan3D outputs were marching cubes meshes, which produced dense but artistically messy triangle soup. To address this the team released Hunyuan3D PolyGen in July 2025.[19][27] PolyGen treats meshes as token sequences and generates them autoregressively using a custom Blocked and Patchified Tokenization scheme that the team claims compresses tokens per face by roughly 74 percent.[19] The model supports both triangle and quadrilateral topology and can generate meshes with more than 20,000 faces while preserving clean edge flow suitable for animation and rigging. Tencent reports that internal artists using PolyGen reduced average modelling time from eight hours to two and a half hours per asset.[19]
Controllable generation
Hunyuan3D Omni, announced in September 2025, sits on top of the 2.1 base and accepts four additional conditioning signals: skeletal pose, sparse or dense point clouds, voxel grids, and 3D bounding boxes.[7][28] Tencent describes it as the ControlNet of 3D.[28] The system uses a lightweight unified control encoder rather than a separate head per modality, and a difficulty aware training curriculum that progressively introduces harder conditioning combinations.[7]
World generation
The HunyuanWorld branch generates whole scenes rather than single objects. HunyuanWorld 1.0 produces a 360 degree panoramic proxy of a scene from a sentence or an image, then converts it into an explorable mesh with disentangled object layers.[11][6] The output is compatible with standard computer graphics pipelines and is intended for virtual reality and game development. A lighter HunyuanWorld 1.0 Lite variant followed for resource constrained deployments, and HunyuanWorld 1.5 in December 2025 added real time interactive playback under the name WorldPlay.[22] HunyuanWorld 2.0 (HY-World-2.0) shipped on 16 April 2026.[6]
What can Hunyuan 3D do?
The table below summarises the input modes, outputs, and feature additions reported across the major versions.
| Capability | 1.0 | 2.0 | 2.1 | 2.5 | 3.0 |
|---|---|---|---|---|---|
| Text to 3D | Yes | Yes | Yes | Yes | Yes |
| Image to 3D | Yes | Yes | Yes | Yes | Yes |
| Multi view image input | No | Partial | Yes | Yes | Yes (up to four views) |
| Sketch to 3D | No | No | Limited | Yes | Yes |
| PBR texture output | No | No | Yes | Yes (4K) | Yes (4K with bump) |
| Geometric resolution | n/a | Not stated | Not stated | 1,024 cubed | 1,536 cubed |
| Quad mesh topology | No | No | Via PolyGen | Via PolyGen | Native and PolyGen |
| Skeletal rigging | No | No | No | Optimized skinning | Yes via Hunyuan 3D Studio |
| Output formats | OBJ | OBJ, GLB | OBJ, GLB | OBJ, GLB | OBJ, GLB |
| Engine compatibility | Generic | Blender addon | Blender, Unity | Blender, Unity, Unreal | Blender, Unity, Unreal |
| End to end production pipeline | No | No | No | Partial | Yes (Studio) |
Tencent reports that Hunyuan 3D 3.0 roughly triples the modeling precision of earlier versions.[16][21] A CLIP score of 0.821, with 15 percent better geometric precision and 20 percent better texture fidelity than Tripo 2, was reported for the earlier 2.5 model in a third-party summary.[20] These numbers come from Tencent's own evaluations and have not been independently reproduced at the time of writing.
Is Hunyuan 3D open source?
Hunyuan3D weights are not released under a standard permissive licence. Instead Tencent uses a bespoke Tencent Hunyuan 3D Community License Agreement, first published with version 2.0 on 21 January 2025 and updated for 2.1 in June 2025.[9][10] The licence grants a non exclusive, non transferable, royalty free right to use, reproduce, modify, and distribute the materials within a defined territory. That territory is the world, with the express exclusion of the European Union, the United Kingdom, and South Korea.[9] Users in those three jurisdictions are not licensed to use, modify, or distribute the model under the community licence.[9]
Hunyuan3D 2.1 marked the turning point on openness. Its technical report calls it "the first fully open-sourced solution for PBR-textured 3D asset generation," arguing that the release "bridges the gap between academic research and scalable content creation."[31] It was the first version to release the full training code in addition to the weights, which Tencent describes as fully open source within the licensed territory.[4][31]
The agreement also restricts the use of model outputs to improve other AI models, prohibits military applications, and includes the usual indemnity and disclaimer clauses.[9] Licensees whose products had more than 1 million monthly active users on the release date must request a separate licence from Tencent.[9][10] The territory restriction is the most contentious element. A Hacker News thread on the 2.0 release flagged the exclusion, with one commenter asking whether it was tied to EU AI regulation, and Tencent kept the same carve out in the 2.1 licence.[30][10]
Not every component of the project ships under the same terms. Hunyuan3D 1.0 itself was released under the Tencent Hunyuan Non-Commercial License Agreement, and its Dust3R based baking module was provided under Creative Commons BY NC SA 4.0, which is also non commercial.[5] Subsequent releases avoided that dependency.
How much does the Hunyuan 3D API cost?
The global Hunyuan 3D creation engine is offered through two channels. The first is a consumer web platform at the Hunyuan 3D portal, which gives every signed in user 20 free generations per day.[1] The second is an enterprise API on Tencent Cloud, available in English and aimed at integration into existing applications. New enterprise accounts receive 200 free credits at registration.[14]
Beyond the free tier the API uses a prepaid credit model. A Professional Edition call that generates a 3D model from text or image with ordinary texture mode consumes 25 credits, with additional options raising the cost.[13] An Express Edition call in the same mode consumes 15 credits.[13] Credit packs are valid for one year from purchase and unused credits expire at the end of that window.[14] The settlement order applies prepaid packs before falling back to pay as you go billing.[14] Tencent Cloud's English billing page lists prepaid packs from 1,000 credits for USD 15 (USD 0.015 per credit) to 100,000 credits for USD 1,350, and postpaid usage at USD 0.02 per credit.[13]
Output formats include OBJ and GLB, and Tencent advertises direct compatibility with Unity, Unreal Engine, and Blender. Specific platform integrations include an official Blender addon released in January 2025 and ComfyUI Partner Nodes for Hunyuan 3D 3.0, announced by ComfyUI in February 2026.[29]
How does Hunyuan 3D compare to Tripo, Meshy, and Rodin?
The table below compares Hunyuan 3D 3.0 with three commonly cited rivals in the generative 3D space as of May 2026. Figures come from each vendor's public documentation and from independent reviews where cited; capabilities not confirmed by primary sources are left blank.
| System | Vendor | Open weights | PBR textures | Geometric resolution | Notable strength |
|---|---|---|---|---|---|
| Hunyuan 3D 3.0 | Tencent | No (API and web platform; open weights up to 2.1 under a community licence) | Yes, 4K | 1,536 cubed | Hard surface geometry, scale, open weights for earlier versions |
| Tripo P1 | Tripo AI | No | Yes | Not disclosed | Organic shapes, generation speed |
| Meshy 6 | Meshy | No | Yes | Not disclosed | Stylized character art, mature web product |
| Rodin Gen-2 | Deemos | No | Yes | Not disclosed | High end realism, sculpt quality |
In head to head reviews from outlets such as 3D AI Studio, Vset3D, and the Scenario knowledge base, reviewers tend to converge on a pattern.[24][25] Hunyuan is favoured for hard surface modelling and for the price of admission, since the weights are free to download.[25] Tripo is praised for organic creatures and stylised content and for fast iteration. Meshy is rated highly for its product polish, documentation, and security certifications including SOC 2 and ISO 27001. Rodin Gen 2 is positioned as the realism leader for high budget productions. These assessments are inevitably subjective and depend heavily on the prompts and reference images used.
The central differentiator is that Hunyuan 3D is the only family in this group with downloadable open weights (up to version 2.1), while Tripo, Meshy, and Rodin are closed hosted products. That makes Hunyuan the default choice for teams that need local inference, on premise deployment, or freedom from per asset licensing fees within the permitted territory.
Reception
The response to Hunyuan 3D inside the open source community has been broadly positive. Tencent reported more than three million Hugging Face downloads at the global launch and calls the series "one of the world's most popular 3D open-source models."[1] ComfyUI added Hunyuan 3D 3.0 as API-backed Partner Nodes in February 2026, and an official Blender addon has been available since January 2025.[29][3]
Criticism has focused on three areas. The first is the territory carve out in the community licence, which prevents legal commercial use in the European Union, the United Kingdom, and South Korea and frustrates contributors based in those regions.[9][30] The second is hardware cost. Running the 2.1 combined pipeline requires roughly 29 GB of VRAM, beyond most consumer GPUs without offloading or quantization, and the larger 2.5 and 3.0 models are not available for local inference at all.[4] A community fork named Hunyuan3D 2GP appeared to address low VRAM environments. The third is mesh quality at the low end. Reviewers consistently noted that the raw marching cubes output before PolyGen was unsuitable for animation.[19]
Academic reception has tracked the open releases closely. The Hunyuan3D 2.1 technical report appeared on arXiv as 2506.15442 in June 2025, the HunyuanWorld 1.0 technical report as 2507.21809 in July 2025, and the Hunyuan 3D Studio paper as 2509.12815 in September 2025.[31][11][12] All have been cited in subsequent work on 3D scene generation and end to end 3D asset pipelines.
Industrial adoption has been led by Tencent's own game studios and by enterprise customers on Tencent Cloud: Tencent says more than 150 enterprises in Mainland China, including Unity China, the consumer 3D printer maker Bambu Lab and the AI content platform Liblib, had integrated the model before the global launch.[1][15] Outside Tencent's direct partnerships, the model has also been picked up by independent 3D printing services, e commerce visualization startups, and indie game developers who appreciate the open weights and the absence of per asset licensing fees within the permitted territory.
Continue reading
- TencentTencent Holdings Limited (Chinese: 腾讯控股有限公司; pinyin: Téngxùn) is a Chinese multinational technology and entertainment conglomerate based in Shenzhen, Guangdong, and is one of the world's largest internet…Mentioned 33 times above
- HunyuanHunyuan is Tencent's brand for its foundation models, an umbrella covering large language models, video generation, image synthesis, 3D asset creation, and multimodal reasoning.Mentioned 32 times above
- HunyuanVideoHunyuanVideo is an open-source video generation model developed by Tencent and released on December 3, 2024, with over 13 billion parametersMentioned twice above
- Diffusion modelA diffusion model is a generative model that learns to transform samples from a simple reference distribution into samples resembling a data distribution by reversing a gradual corruption process.Listed in See also
- Tripo P1Tripo P1, marketed in full as Tripo Smart Mesh P1.0, is a production grade native 3D diffusion model that generates clean, engine ready 3D meshes from text or image prompts in as little as two seconds.Listed in See also
- Meshy 6Meshy 6 is the sixth major release of Meshy AI's generative 3D generation platform, a hosted generative AI service that turns text prompts or images into textured 3D models (meshes) for games, animation, and…Listed in See also
More from See also
More to read
References
- ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8Tencent. "Tencent Announces Global Launch of Hunyuan 3D Engine to Empower Creators with Advanced Creation Tools." Press release, 25 November 2025. tencent.com/...2202235
- ^PR Newswire. "Tencent Announces Global Launch of Hunyuan 3D Engine." 25 November 2025. prnewswire.com/...dvanced-creation-tools-302626280
- ^1 ^2 ^3 ^4 ^5 ^6Tencent Hunyuan. "Hunyuan3D-2 GitHub repository." github.com/...Hunyuan3D-2
- ^1 ^2 ^3 ^4 ^5Tencent Hunyuan. "Hunyuan3D-2.1 GitHub repository." github.com/...Hunyuan3D-2.1
- ^1 ^2 ^3Tencent Hunyuan. "Hunyuan3D-1 GitHub repository." github.com/...Hunyuan3D-1
- ^1 ^2Tencent Hunyuan. "HunyuanWorld-1.0 GitHub repository." github.com/...HunyuanWorld-1.0
- ^1 ^2Tencent Hunyuan. "Hunyuan3D-Omni GitHub repository." github.com/...Hunyuan3D-Omni
- ^1 ^2Hugging Face. "tencent/Hunyuan3D-2 model card." huggingface.co/...Hunyuan3D-2
- ^1 ^2 ^3 ^4 ^5 ^6 ^7Tencent Hunyuan. "Hunyuan3D 2.0 Community License Agreement." 21 January 2025. github.com/...LICENSE
- ^1 ^2 ^3Tencent Hunyuan. "Hunyuan3D 2.1 Community License Agreement." 13 June 2025. github.com/...LICENSE
- ^1 ^2arXiv. "HunyuanWorld 1.0: Generating Immersive, Explorable, and Interactive 3D Worlds from Words or Pixels." 2507.21809, July 2025. arxiv.org/...2507.21809
- ^1 ^2arXiv. "Hunyuan3D Studio: End to End AI Pipeline for Game Ready 3D Asset Generation." 2509.12815, September 2025. arxiv.org/...2509.12815v1
- ^1 ^2 ^3Tencent Cloud. "Tencent Hunyuan 3D Billing Overview." tencentcloud.com/...74121
- ^1 ^2 ^3Tencent Cloud. "Tencent Hunyuan 3D Global Purchase Guide." staticintl.cloudcachetci.com/...1281_74120_en.pdf
- ^1 ^2Auganix. "Tencent Launches Hunyuan 3D AI Asset Generation Engine." 26 November 2025. auganix.org/...nyuan-3d-ai-asset-generation-engine
- ^1 ^2 ^3Blue Lightning TV. "Tencent's Hunyuan 3D 3.0 Debuts With Triple Precision, Ultra HD Geometry, and API Access." 20 September 2025. bluelightningtv.com/...-hd-geometry-and-api-access
- ^1 ^2VoxelMatters. "Tencent launches updated Hunyuan 3D 3.0 platform for 3D model generation." September 2025. voxelmatters.com/...atform-for-3d-model-generation
- ^AI Base. "Tencent HunYuan 3D Generative Model Officially Releases Version 2.5." April 2025. aibase.com/...17448
- ^1 ^2 ^3 ^4AI Base. "Tencent Hunyuan Launches the Industry's First Art Level 3D Generation Large Model Hunyuan3D-PolyGen." July 2025. aibase.com/...19515
- ^Vset3D. "Hunyuan 3D 2.5 Tencent Pushes the Boundaries of 3D Generation with AI." vset3d.com/...-boundaries-of-3d-generation-with-ai
- ^Vset3D. "Hunyuan 3D 3.0 The Future of AI Powered Modeling." vset3d.com/...-0-the-future-of-ai-powered-modeling
- ^The Decoder. "Tencent releases Hunyuan World Model 1.0 Lite for faster, resource efficient 3D scene generation." the-decoder.com/...urce-ai-for-3d-scene-generation
- ^Analytics Vidhya. "Tencent Hunyuan3D-1.0: 3D Modeling with AI Driven Speed and Precision." November 2024. analyticsvidhya.com/...hunyuan3d-1-0
- ^3DAI Studio. "Best 3D Model Generation APIs in 2026." 3daistudio.com/...best-3d-model-generation-apis-2026
- ^1 ^2Medium (Sam Glassenberg). "Head to Head AI image to 3D comparison: Hunyuan vs Meshy vs Tripo." medium.com/...son-hunyuan-vs-meshy-vs-f99cb38faa39
- Tencent Cloud (X). "We are officially taking Hunyuan 3D Engine global." 25 November 2025. x.com/...1993986997468156084
- ^Tencent HY (X). "Introducing Hunyuan3D-PolyGen." 7 July 2025. x.com/...1942174221976981881
- ^1 ^2Tencent HY (X). "Introducing Tencent Hunyuan3D-Omni." September 2025. x.com/...1971495031040283125
- ^1 ^2ComfyUI Blog. "Hunyuan 3D 3.0 in ComfyUI: State of the Art 3D Asset Generation." blog.comfy.org/...hunyuan-3d-30-in-comfyui-state-of
- ^1 ^2Hacker News discussion. "Hunyuan 3D 2.0 community licence territory restrictions." news.ycombinator.com/item
- ^1 ^2 ^3 ^4 ^5arXiv. "Hunyuan3D 2.1: From Images to High-Fidelity 3D Assets with Production-Ready PBR Material." 2506.15442, June 2025. arxiv.org/...2506.15442v1
- ^Tencent Cloud. "SubmitHunyuanTo3DProJob" API documentation (Tencent HY 3D Global). intl.cloud.tencent.com/...75540
Improve this article
Add missing citations, update stale details, or write a clearer explanation. Anyone can edit: changes go live when you publish, and every edit is checked by an editor.
8 revisions · v8 · 3,380 words · full history
Fact-checks are independent of edits: a reviewer re-verifies the article against its sources and stamps the date. How we verify
Research and drafting on this wiki are AI-assisted, under named human editorial standards. How AI is used here
Reviewer note: Independent fact-check of the full article against its cited sources (110 claims, 33 sources fetched); 12 material + 11 minor fixed
Cite this page: AI Wiki. "Hunyuan 3D." aiwiki.ai, updated 30 Oct 2026, fact-checked 5 Nov 2026. CC BY 4.0. https://aiwiki.ai/wiki/hunyuan_3d


