60K 32x32 color images in 100 fine-grained classes grouped into 20 superclasses. More challenging than CIFAR-10.
Accuracy is the reported evaluation metric for CIFAR-100. Codesota tracks published model scores on this metric so readers can compare state-of-the-art results across sources and model families.
Higher is better
Muted rows were not state of the art when published — an earlier or same-year result already scored better.
| Rank | Model | Trust | Score | Year | Links | Fix |
|---|---|---|---|---|---|---|
| 01 | EVA-02-L | vendor | 97.15 | 2026 | Source ↗ | Looks wrong? |
| 02 | CoAtNet-7 | vendor | 96.38 | 2026 | Source ↗ | Looks wrong? |
| 03 | ConvNeXt V2-H | vendor | 96.17 | 2026 | Source ↗ | Looks wrong? |
| 04 | MAE ViT-H/14 | vendor | 96.08 | 2026 | Source ↗ | Looks wrong? |
| 05 | SwinV2-G | vendor | 96.01 | 2026 | Source ↗ | Looks wrong? |
| 06 | DeiT III-H/14 | vendor | 95.94 | 2026 | Source ↗ | Looks wrong? |
| 07 | InternImage-XL | vendor | 95.77 | 2026 | Source ↗ | Looks wrong? |
| 08 | FasterViT-6 | vendor | 95.72 | 2026 | Source ↗ | Looks wrong? |
| 09 | vit-h-14 | vendor | 94.55 | 2026 | Source ↗ | Looks wrong? |
| 10 | AIMv2-3B | verified | 94.5 | 2026 | Source ↗ | Looks wrong? |
| 11 | AIMv2-1B | verified | 94.1 | 2026 | Source ↗ | Looks wrong? |
| 12 | ViT-L/16 (IN-21K) | verified | 93.25 | 2026 | Source ↗ | Looks wrong? |
| 13 | efficientnet-b7 | vendor | 91.7 | 2026 | Source ↗ | Looks wrong? |
| 14 | vit-b-16 | vendor | 91.48 | 2026 | Source ↗ | Looks wrong? |
| 15 | resnet-50 | vendor | 78.04 | 2026 | Source ↗ | Looks wrong? |