# Zero-shot Image Classification on ImageNet 1k and Six Robustness Variants (test)

Source: [Zero-shot Image Classification on ImageNet 1k and Six Robustness Variants (test)](<https://www.sota2.com/research/sota/zero-shot-image-classification-on-imagenet-1k-and-six-robustness-variants-test>)

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

Data updated: 2026-02-26T06:09:06.310Z

Evaluation of zero-shot image classification performance on the standard ImageNet 1k test set combined with six specified robustness variants.

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 | Zero-shot Image Classification |
| Dataset (including any stated split) | ImageNet 1k and six robustness variants (test) |
| Metric | Zero-shot Mean Accuracy |
| Unit | PERCENT |
| Direction | Higher 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

- [Zero-shot Mean Accuracy](<https://www.sota2.com/research/sota/zero-shot-image-classification-on-imagenet-1k-and-six-robustness-variants-test.md?metric=Zero-shot%20Mean%20Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Zero-shot Mean Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| TuneCLIP | 65.58 | Not recorded | 6969c607e25ad2fa9d1ec906 |
| Baseline | 63.12 | Not recorded | 6969c607e25ad2fa9d1ec906 |
| TuneCLIP | 59.36 | Not recorded | 6969c607e25ad2fa9d1ec906 |
| Baseline | 57.67 | Not recorded | 6969c607e25ad2fa9d1ec906 |
| OpenCLIP | 54.99 | Not recorded | 6969c607e25ad2fa9d1ec906 |
| FastCLIP | 54.57 | Not recorded | 6969c607e25ad2fa9d1ec906 |
| OpenCLIP | 40.21 | Not recorded | 6969c607e25ad2fa9d1ec906 |
| FastCLIP | 39.22 | Not recorded | 6969c607e25ad2fa9d1ec906 |

### TuneCLIP

Zero-shot Mean Accuracy: 65.58. Result ID: 6997bb8e5698665ec4121b15.

[Reporting paper](<https://www.sota2.com/research/paper/6969c607e25ad2fa9d1ec906.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6969c607e25ad2fa9d1ec906.md>)

Recorded evaluation configuration: {"Base Model":"SigLIP ViT-B/16","WCT. (hrs)":9.27}. This record does not assert independent reproduction.

### Baseline

Zero-shot Mean Accuracy: 63.12. Result ID: 6997bb8e5698665ec4121b12.

[Reporting paper](<https://www.sota2.com/research/paper/6969c607e25ad2fa9d1ec906.md>)

Recorded evaluation configuration: {"Base Model":"SigLIP ViT-B/16","WCT. (hrs)":"N/A"}. This record does not assert independent reproduction.

### TuneCLIP

Zero-shot Mean Accuracy: 59.36. Result ID: 6997bb8e5698665ec4121b11.

[Reporting paper](<https://www.sota2.com/research/paper/6969c607e25ad2fa9d1ec906.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6969c607e25ad2fa9d1ec906.md>)

Recorded evaluation configuration: {"Base Model":"OpenAI ViT-B/16","WCT. (hrs)":8.62}. This record does not assert independent reproduction.

### Baseline

Zero-shot Mean Accuracy: 57.67. Result ID: 6997bb8e5698665ec4121b0e.

[Reporting paper](<https://www.sota2.com/research/paper/6969c607e25ad2fa9d1ec906.md>)

Recorded evaluation configuration: {"Base Model":"OpenAI ViT-B/16","WCT. (hrs)":"N/A"}. This record does not assert independent reproduction.

### OpenCLIP

Zero-shot Mean Accuracy: 54.99. Result ID: 6997bb8e5698665ec4121b10.

[Reporting paper](<https://www.sota2.com/research/paper/6969c607e25ad2fa9d1ec906.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74261e209131cea6003e.md>)

[Implementation](<https://github.com/LAION-AI/scaling-laws-openclip>)

Recorded evaluation configuration: {"Base Model":"OpenAI ViT-B/16","WCT. (hrs)":5.46}. This record does not assert independent reproduction.

### FastCLIP

Zero-shot Mean Accuracy: 54.57. Result ID: 6997bb8e5698665ec4121b0f.

[Reporting paper](<https://www.sota2.com/research/paper/6969c607e25ad2fa9d1ec906.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75001e209131cea6526b.md>)

Recorded evaluation configuration: {"Base Model":"OpenAI ViT-B/16","WCT. (hrs)":4.21}. This record does not assert independent reproduction.

### OpenCLIP

Zero-shot Mean Accuracy: 40.21. Result ID: 6997bb8e5698665ec4121b14.

[Reporting paper](<https://www.sota2.com/research/paper/6969c607e25ad2fa9d1ec906.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74261e209131cea6003e.md>)

[Implementation](<https://github.com/LAION-AI/scaling-laws-openclip>)

Recorded evaluation configuration: {"Base Model":"SigLIP ViT-B/16","WCT. (hrs)":7.55}. This record does not assert independent reproduction.

### FastCLIP

Zero-shot Mean Accuracy: 39.22. Result ID: 6997bb8e5698665ec4121b13.

[Reporting paper](<https://www.sota2.com/research/paper/6969c607e25ad2fa9d1ec906.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75001e209131cea6526b.md>)

Recorded evaluation configuration: {"Base Model":"SigLIP ViT-B/16","WCT. (hrs)":4.28}. This record does not assert independent reproduction.

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