whatmodel.sh / image / PixArt-α (PixArt-XL-2-1024-MS)
PixArt-α (PixArt-XL-2-1024-MS)
↗ share
+ contribute
PixArt-α (PixArt-XL-2-1024-MS) is an open-weights image model from PixArt (Huawei Noah's Ark Lab et al.), released November 2023. Its published size is 0.6B parameters. Licence: openrail++.
Model information sources
Release date ↗ · Weights licence ↗ · Parameter count ↗
Model-zoo row for exactly this checkpoint; the HF card's table also lists 'PixArt-α | Diff | 0.6B'. Covers the diffusion transformer only; T5-XXL text encoder and SD VAE not included.
Score within the LMArena text-to-image comparison set only.
$ how it compares
The image models ranked either side of PixArt-α (PixArt-XL-2-1024-MS), out of 197 with a score.
| # | model | score | where |
|---|---|---|---|
| 161 |  SSD-1B | 36±4 | local |
| 162 |  JanusFlow-1.3B | 36±6 | comparison only |
| 163 | PixArt-α (PixArt-XL-2-1024-MS) | 36±2 | local |
| 164 |  Emu3 | 36±2 | local |
| 165 |  Ovis-U1-3B | 35±6 | comparison only |
$ what it's good at
text to image▂▅▅▃▄▄█▃▁▂▁▁ #97 of 126 · 69% of its score · 3 tests
Beats Latent Diffusion text2img-large (LDM-KL-8-G, LAION-400M), VQ-Diffusion (ITHQ), Stable Diffusion XL 0.9 (base) here, though they rank higher overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| GenAI-Arena text-to-image (Elo, 4 Jun 2024) | 1052.0 ✎ | ▃▃··▃·▃·█·▃▃ beats 4/8 | 34% |
| ConceptMix (k=2 full-mark score) | 37.0% ✎ | ▄·▄·█▄▄▄···▄ beats 3/7 | 26% |
| WISE (WiScore overall, GPT-4o judge) | 47.0% ✎ | ▄·▆▂▆█▄▆▂▄·▄ beats 12/22 | 8.9% |
other tests1 not in a task · 29% of its score
| test | result | vs the others in its run | counts |
|---|---|---|---|
| EvalAlign human text-image alignment score | 5.1100 mean human alignment score ✎ | █▄▄·█·▄█▄▄·▄ beats 8/11 | 29% |
$ where to run it
Other ways to run it
- Run it as published · needs ~13.6 GB, NVIDIA only · from huggingface.co/PixArt-alpha (official)
uv run --with torch --with diffusers --with transformers --with accelerate --with safetensors --with sentencepiece python -c "import torch; from diffusers import PixArtAlphaPipeline; p=PixArtAlphaPipeline.from_pretrained('PixArt-alpha/PixArt-XL-2-1024-MS', torch_dtype=torch.float16).to('cuda'); p(prompt='An astronaut riding a green horse').images[0].save('out.png')"
Official weights and documentation: huggingface.co/PixArt-alpha ↗
$ how the score was worked out · GenAI Arena: An Open Evaluati… 34%, EvalAlign: Supervised Fine-Tu… 29%, ConceptMix: A Compositional I… 26% and 1 more
Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works
| source | results | share of the score |
|---|---|---|
 GenAI Arena: An Open Evaluation Platform for Generative Models (arXiv v3) | 1 | 34% |
| EvalAlign: Supervised Fine-Tuning Multimodal LLMs with Human-Aligned Data for Evaluating Text-to-Image Models (arXiv:2406.16562v3) | 1 | 29% |
| ConceptMix: A Compositional Image Generation Benchmark with Controllable Difficulty (arXiv v1) | 1 | 26% |
| WISE: A World Knowledge-Informed Semantic Evaluation for Text-to-Image Generation (arXiv v4) | 1 | 9% |
| prior · pull toward the middle | 2% | |
| weighted average | = 35.7 |
$ improve this page
Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.
data updated · sources and licences · json