# Class-conditional image synthesis on ImageNet 256x256 1.0 (val)

Source: [Class-conditional image synthesis on ImageNet 256x256 1.0 (val)](<https://www.sota2.com/research/sota/class-conditional-image-synthesis-on-imagenet-256x256-1-0-val>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-02-26T05:44:15.589Z

Evaluation of class-conditional image synthesis models on the validation split of the ImageNet dataset rendered at 256x256 resolution (version 1.0), typically measured using FID, IS, Precision, Recall, and CAS metrics.

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

| Scope | Recorded value |
| --- | --- |
| Task | Class-conditional image synthesis |
| Dataset (including any stated split) | ImageNet 256x256 1.0 (val) |
| Metric | FID |
| Unit | SCALAR |
| Direction | Lower is better |
| Results loaded | 8 of 8 |
| Coverage | All results currently returned for this benchmark |

Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol.

## Metrics

- [FID](<https://www.sota2.com/research/sota/class-conditional-image-synthesis-on-imagenet-256x256-1-0-val.md?metric=FID>): Lower is better; unit: SCALAR.

- [IS](<https://www.sota2.com/research/sota/class-conditional-image-synthesis-on-imagenet-256x256-1-0-val.md?metric=IS>): Higher is better; unit: SCALAR.

- [Precision](<https://www.sota2.com/research/sota/class-conditional-image-synthesis-on-imagenet-256x256-1-0-val.md?metric=Precision>): Higher is better; unit: PERCENT.

- [Recall](<https://www.sota2.com/research/sota/class-conditional-image-synthesis-on-imagenet-256x256-1-0-val.md?metric=Recall>): Higher is better; unit: PERCENT.

- [CAS Top-1 Accuracy](<https://www.sota2.com/research/sota/class-conditional-image-synthesis-on-imagenet-256x256-1-0-val.md?metric=CAS%20Top-1%20Accuracy>): Higher is better; unit: PERCENT.

- [CAS Top-5 Accuracy](<https://www.sota2.com/research/sota/class-conditional-image-synthesis-on-imagenet-256x256-1-0-val.md?metric=CAS%20Top-5%20Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | FID | Result date | Report paper |
| --- | --- | --- | --- |
| MaskGIT | 6.18 | Not recorded | 68fa6bc61e209131cea303c7 |
| BigGAN-deep | 6.95 | Not recorded | 68fa6bc61e209131cea303c7 |
| ADM | 10.94 | Not recorded | 68fa6bc61e209131cea303c7 |
| Improved DDPM | 12.26 | Not recorded | 68fa6bc61e209131cea303c7 |
| VQGAN | 15.78 | Not recorded | 68fa6bc61e209131cea303c7 |
| VQGAN\* | 18.65 | Not recorded | 68fa6bc61e209131cea303c7 |
| VQVAE-2 | 31.11 | Not recorded | 68fa6bc61e209131cea303c7 |
| DCTransformer | 36.51 | Not recorded | 68fa6bc61e209131cea303c7 |

### MaskGIT

FID: 6.18. Result ID: 698ea2fdffc3901de09f589d.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6bc61e209131cea303c7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6bc61e209131cea303c7.md>)

Recorded evaluation configuration: {"# params":"227M","# steps":"8"}. This record does not assert independent reproduction.

### BigGAN-deep

FID: 6.95. Result ID: 698ea2fdffc3901de09f5897.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6bc61e209131cea303c7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa65661e209131cea0a090.md>)

[Implementation](<https://tfhub.dev/s?q=biggan>)

Recorded evaluation configuration: {"# params":"160M","# steps":"1"}. This record does not assert independent reproduction.

### ADM

FID: 10.94. Result ID: 698ea2fdffc3901de09f5899.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6bc61e209131cea303c7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6a0e1e209131cea25dba.md>)

[Implementation](<https://github.com/openai/guided-diffusion>)

Recorded evaluation configuration: {"# params":"554M","# steps":"250"}. This record does not assert independent reproduction.

### Improved DDPM

FID: 12.26. Result ID: 698ea2fdffc3901de09f5898.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6bc61e209131cea303c7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa694d1e209131cea215bb.md>)

[Implementation](<https://github.com/openai/improved-diffusion>)

Recorded evaluation configuration: {"# params":"280M","# steps":"250"}. This record does not assert independent reproduction.

### VQGAN

FID: 15.78. Result ID: 698ea2fdffc3901de09f589b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6bc61e209131cea303c7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6a371e209131cea26a4d.md>)

[Implementation](<https://git.io/JnyvK>)

Recorded evaluation configuration: {"# params":"1.4B","# steps":"256"}. This record does not assert independent reproduction.

### VQGAN\*

FID: 18.65. Result ID: 698ea2fdffc3901de09f589c.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6bc61e209131cea303c7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6a371e209131cea26a4d.md>)

[Implementation](<https://git.io/JnyvK>)

Recorded evaluation configuration: {"# params":"227M","# steps":"256","setup":"same architecture and setup as MaskGIT"}. This record does not assert independent reproduction.

### VQVAE-2

FID: 31.11. Result ID: 698ea2fdffc3901de09f589a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6bc61e209131cea303c7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa65d61e209131cea0c943.md>)

Recorded evaluation configuration: {"# params":"13.5B","# steps":"5120"}. This record does not assert independent reproduction.

### DCTransformer

FID: 36.51. Result ID: 698ea2fdffc3901de09f5896.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6bc61e209131cea303c7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa696e1e209131cea21fd2.md>)

Recorded evaluation configuration: {"# params":"738M","# steps":"\>1024"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.