whatmodel.sh / image / UNO
UNO
↗ share
+ contribute
UNO is an open-weights image model from ByteDance, released April 2025. Licence: cc-by-nc-4.0.
Catalogue identity notes
UNO (Wu et al., 'Less-to-More Generalization: Unlocking More Controllability by In-Context Generation', arXiv 2504.02160; ByteDance Intelligent Creation Team), a subject-driven multi-reference generator trained from FLUX.1-dev and released in April 2025 as a single LoRA checkpoint (bytedance-research/UNO, dit_lora.safetensors). It is the only UNO release; OmniGen2 report Table 4 cites it as 'UNO [97]' and ImagenWorld Table 2 as 'UNO (Wu et al., 2025c)'. Parameter count not stated by the makers (LoRA on the 12B FLUX.1-dev transformer), so params_b is left out. Licence: the card's front matter says apache-2.0, but its text says 'Our code is released under the Apache 2.0 License, while our models are under the CC BY-NC 4.0 License. Any models related to FLUX.1-dev base model must adhere to the original licensing terms.' Recorded as the stricter weights licence (cc-by-nc-4.0); the base model FLUX.1-dev is under the FLUX.1 [dev] non-commercial licence.
Model information sources
Score within the LMArena text-to-image comparison set only.
$ how it compares
The image models ranked either side of UNO, out of 197 with a score, then more from ByteDance.
| # | model | score | where | ||
|---|---|---|---|---|---|
| 127 |  FLUX.1 [dev] | 44±1 | api $0.025/image +local | ||
| 128 |  UniPic2-SD3.5M-Kontext-2B | 44±6 | comparison only | ||
| 129 | UNO | 43±4 | local | ||
| 130 |  ChronoEdit-14B | 43±3 | local | ||
| 131 |  Lumina-Image 2.0 | 43±4 | api $0.075/image +local | ||
| more from ByteDance | |||||
| 30 | Seedream 4.5 | 64±1 | api $0.040/image | ||
| 156 | SDXL-Lightning (4-step UNet) | 37±3 | local | ||
$ what it's good at
text to image▂▅▅▃▄▄█▃▁▂▁▁ #79 of 126 * · 79% of its score · 1 test
Beats Latent Diffusion text2img-large (LDM-KL-8-G, LAION-400M), VQ-Diffusion (ITHQ), FLUX.1 Kontext [dev] here, though they rank higher overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| ImagenWorld text-guided generation (human overall) | 78.0% ✎ | ▄··█▄·▄··▄·▄ beats 3/6 | 79% |
image to image / editing▁▁·▃▂▃█▃▃▄▁▁ #58 of 62 * · 14% of its score · 1 test
Trails Stable Diffusion 3.5 Large, OmniGen2, BAGEL here, though they rank lower overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| OmniContext average (GPT-4.1 judge) | 47.1% ✎ | ▃▃▃··▃·····█ beats 2/6 | 14% |
$ where to run it
Other ways to run it
- Run it as published · needs ~16 GB, NVIDIA only · from github.com/bytedance (official)
git clone https://github.com/bytedance/UNO.git && cd UNO && pip install -e . && python inference.py --prompt "A clock on the beach is under a red sun umbrella" --image_paths "assets/clock.png" --width 704 --height 704 --offload --model_type flux-dev-fp8
Weights and documentation: huggingface.co/bytedance-research ↗
$ how the score was worked out · ImagenWorld: Stress-Testing I… 79%, BAAI — OmniGen2 technical rep… 14%
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 |
|---|---|---|
 ImagenWorld: Stress-Testing Image Generation Models with Explainable Human Evaluation on Open-ended Real-World Tasks | 1 | 79% |
| BAAI — OmniGen2 technical report | 1 | 14% |
| prior · pull toward the middle | 7% | |
| weighted average | = 43.2 |
$ 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