PixArt Models by Huawei Noah's Ark Lab
1 model2024
Last refreshed 2026-05-19. Next refresh: weekly.
Details
ResearcherHuawei Noah's Ark Lab
LicenseApache 2.0OSI-approved
Commercial useCommercial use: permitted
Models1
Released2024
Capabilities
MultimodalAll models
About
Huawei Noah's Ark Lab's PixArt family of diffusion transformer text-to-image models. Known for efficient training and high-quality output — PixArt-α introduced fast DiT training, PixArt-Σ extended to 4K resolution generation with 'weak-to-strong training'.
Decision facts
Current Variants
Use-when guidance is based on each model's tracked capabilities, context window, release date, and replacement status.
1 in view
PixArt-ΣCurrent
Use when the workload needs image generation, 600M parameters, and multimodal inputs.
2024-03image generation600M parametersmultimodal inputs
| Model | Use when | Released | Signals | Status |
|---|---|---|---|---|
| PixArt-Σ | Use when the workload needs image generation, 600M parameters, and multimodal inputs. | 2024-03 | image generation600M parametersmultimodal inputs | Current |
Release Timeline
1 release group2024-03
1 current
PixArt-Σ
Currentimage generation600M parametersmultimodal inputs
Specifications(1 models)
| Model | Released | Parameters | Multimodal |
|---|---|---|---|
| PixArt-Σ | 2024-03 | 600M | Yes |
