whatmodel.sh / image / AnyEdit (AnySD)
AnyEdit (AnySD)
↗ share
+ contribute
AnyEdit (AnySD) is an open-weights image model from Zhejiang University / NUS (AnyEdit), released February 2025. Licence: apache-2.0.
Catalogue identity notes
AnyEdit (Yu et al., CVPR 2025) released as the AnySD editing model (WeiChow/AnySD, pinned cafe016, created 2025-02-20, Apache-2.0); the card notes the public model was retrained on the cleaned AnyEdit data, so its numbers differ slightly from the paper. EdiVal pins 'AnyEdit: repository at https://github.com/weichow23/AnySD/tree/9e7d36e... with edit_type set to general'. Edit-Compass ('AnyEdit Yu et al. (2025)') and WEAVE ('AnyEdit [86] 1B') name the same single public release. WEAVE's first author Wei Chow is an AnyEdit author, so the WEAVE row is by: lab. Comparison-only.
Model information sources
Score within the LMArena text-to-image comparison set only.
$ how it compares
The image models ranked either side of AnyEdit (AnySD), out of 197 with a score.
| # | model | score | where |
|---|---|---|---|
| 177 |  IF-M | 31±3 | local |
| 178 |  Dreamlike Photoreal 2.0 | 30±3 | local |
| 179 | AnyEdit (AnySD) | 30±4 | comparison only |
| 180 |  Stable Diffusion v2.1 | 29±2 | comparison only |
| 181 | Stable Diffusion v2.0 | 29±3 | comparison only |
$ what it's good at
image to image / editing▁▃▁▃▂▃█▃▃▃▁▁ #79 of 81 * · 51% of its score · 1 test
| test | result | vs the others in its run | counts |
|---|---|---|---|
| Edit-Compass (EN, overall) | 26.2% ✎ | ▃▃▂·▃▆▂█·▃·▃ beats 2/20 | 51% |
other tests2 not in a task · 44% of its score
| test | result | vs the others in its run | counts |
|---|---|---|---|
| WEAVEBench in-context multi-turn editing (Avg) lab | 47.2% ✎ | █▄▄·▄▄▄▄█·▄█ beats 3/12 | 25% |
| EdiVal multi-turn editing overall (EdiVal-O after turn 3) | 22.5% ✎ | ▂·▂▃·▂·▃▂·▃█ beats 2/14 | 19% |
$ where to run it
No verified API or download recorded yet · add one
Weights and documentation: huggingface.co/WeiChow ↗
$ how the score was worked out · Edit-Compass & EditReward-Com… 51%, WEAVE: Unleashing and Benchma… 25%, EdiVal-Agent: An Object-Centr… 19%
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 |
|---|---|---|
 Edit-Compass & EditReward-Compass: A Unified Benchmark for Image Editing and Reward Modeling | 1 | 51% |
| WEAVE: Unleashing and Benchmarking the In-context Interleaved Comprehension and Generation (arXiv:2511.11434v1) · lab-reported | 1 | 25% |
| EdiVal-Agent: An Object-Centric Framework for Automated, Fine-Grained Evaluation of Multi-Turn Editing (arXiv:2509.13399v3) | 1 | 19% |
| prior · pull toward the middle | 5% | |
| weighted average | = 29.9 |
$ 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