TRELLIS.2
A solid pick for self-hosting and full data control.
Overview
TRELLIS.2 is Microsoft Research's 3D generation model, released mid-December 2025: 4 billion parameters (double the first generation) dedicated to high-fidelity image-to-3D. Its key technical contribution is O-Voxel, a "field-free" sparse voxel representation that breaks free of iso-surface limits (SDF, Flexicubes): the model handles open surfaces (clothing, foliage), non-manifold geometry and enclosed internal structures — exactly where previous 3D generators failed. Performance is concrete: on an H100 GPU, roughly 3 seconds for a 512³-resolution asset, 17 seconds at 1024³, 60 seconds at 1536³ — fully textured, with complete PBR materials (base color, roughness, metallic, opacity, transparency included). Textured mesh → O-Voxel conversion takes under 10 seconds on CPU, and the reverse under 100 ms with CUDA acceleration, making it a realistic pipeline component rather than just a research demo. The model is open source, self-hostable (16GB NVIDIA GPU minimum recommended), and already integrated into the ComfyUI ecosystem via community wrappers. Within weeks it became the reference foundation for generative 3D: work like Stability AI's Arbor builds directly on top of it. For a studio or product team, it's currently the best free entry point into production-grade 3D asset generation.
Skill profile
Not disclosed
Three independent measures
What people prefer, what the model can do, what the market uses. Each source has its biases and none is enough on its own: we show them side by side, never blended.
Strengths
- Has a free tier
- Open-source and self-hostable
Who is it for
- you want to control cost or self-host
- your data must stay in the EU
Ideal use cases
- 3D asset generation for games and real-time engines
- Product prototyping and 3D e-commerce from a photo
- Asset pipelines with full PBR materials
Access & availability
Key specifications
What it costs per month
Estimate from sourced API prices and a hardware tier for self-hosting. Assumptions are editable.
Assumptions: 36-month amortisation, 0.25 €/kWh, 22 days/month, 1 $ = 0.92 €. Excludes labour, software licences, redundancy.
nAIvigate Trust Index
What the vendor offers and documents for professional use in Europe. An undocumented criterion counts as zero and stays visible: the score rises when the vendor publishes, not when we assume.
19/19 criteria documented · 100 % · 12 of them set by rule
See the criteria▾
Trust Index badge
Show this score in your docs, README or security page. It updates automatically.
[](https://naivigate.com/en/model/trellis-2)Data & compliance
What we know, line by line, with source and date. These lines feed the Data and Enterprise families of the Trust Index above. GDPR compliance depends on your processing, not on the model alone.
| EU data residency | Yes |
| Data processing agreement (DPA) | N/A· self-hosted: data never leaves |
| Transfer outside the EU | No |
| Training on your data | No |
| Configurable retention | N/A· self-hosted: data never leaves |
| Zero Data Retention | N/A· self-hosted: data never leaves |
| Self-hostable | Yes |
| Declared hosting | self |
Quality of this card’s data
This score measures the quality of the data on this card: price, context, Elo, licence, documented, sourced and fresh. It measures neither the model nor its compliance: for that, see the Trust Index above.
5/8 documented · 2/8 primary sources
Run it locally
| Quantization | Disk | RAM / VRAM | Typical hardware |
|---|---|---|---|
| Q4 · recommended | 2.3 GB | 5 GB | Any recent PC/Mac |
| Q8 · balanced | 4.3 GB | 7 GB | 16GB PC / M1+ Mac / 8GB GPU |
| FP16 · max quality | 8 GB | 11 GB | 32GB PC / 24GB Mac / RTX 4070 Ti+ |
Estimates for a moderate context. Long contexts need more RAM (KV cache).
Deploy
Copy-ready commands generated from this card. Adjust context length and GPU count to your hardware.
OpenAI-compatible server for production on NVIDIA GPUs.
pip install vllm
vllm serve microsoft/TRELLIS.2-4B \
--max-model-len 4000 \
--tensor-parallel-size 1 \
--dtype autoAdvanced data · for expertsArchitecture, modalities, detailed cost, full benchmarks▾
| Arena Elo | — |
| MMLU | — |
| GPQA | — |
| HumanEval | — |
| SWE-Bench | — |
| MATH | — |