Citation network of scientific papers. 2708 nodes, 5429 edges, 7 classes. Classic GNN benchmark.
Accuracy is the reported evaluation metric for Cora. 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 | TAPE + RevGAT | verified | 92.9 | 2026 | Source ↗ | Looks wrong? |
| 02 | AuGLM (T5-large) | verified | 91.51 | 2026 | Source ↗ | Looks wrong? |
| 03 | ENGINE | verified | 91.48 | 2026 | Source ↗ | Looks wrong? |
| 04 | InstructGLM | verified | 90.77 | 2026 | Source ↗ | Looks wrong? |
| 05 | GLEM + RevGAT | verified | 88.56 | 2026 | Source ↗ | Looks wrong? |
| 06 | GCNLLMEmb | verified | 88.15 | 2026 | Source ↗ | Looks wrong? |
| 07 | LLaGA (Mistral-7B) | verified | 87.55 | 2026 | Source ↗ | Looks wrong? |
| 08 | SDGAT | verified | 85.29 | 2026 | Source ↗ | Looks wrong? |
| 09 | GCN* (tuned) | verified | 85.08 | 2026 | Source ↗ | Looks wrong? |
| 10 | GAT* (tuned) | verified | 84.64 | 2026 | Source ↗ | Looks wrong? |
| 11 | SGFormer | verified | 84.5 | 2026 | Source ↗ | Looks wrong? |
| 12 | GraphSAGE* (tuned) | verified | 84.18 | 2026 | Source ↗ | Looks wrong? |
| 13 | Polynormer | verified | 83.25 | 2026 | Source ↗ | Looks wrong? |
| 14 | GOAT | verified | 83.18 | 2026 | Source ↗ | Looks wrong? |
| 15 | GAT | verified | 83 | 2026 | Source ↗ | Looks wrong? |
| 16 | GraphGPS | verified | 82.84 | 2026 | Source ↗ | Looks wrong? |
| 17 | Exphormer | verified | 82.77 | 2026 | Source ↗ | Looks wrong? |
| 18 | GraphSAGE | verified | 82.68 | 2026 | Source ↗ | Looks wrong? |
| 19 | NodeFormer | verified | 82.2 | 2026 | Source ↗ | Looks wrong? |
| 20 | NAGphormer | verified | 82.12 | 2026 | Source ↗ | Looks wrong? |
| 21 | GCN | verified | 81.5 | 2026 | Source ↗ | Looks wrong? |