Long-horizon memory Question Answering on LongBench
73.7Accuracy
GPT-4o (ceiling)
Updated 2mo ago
Evaluation Results
| Method | Links | |
|---|---|---|
2026.05 | 73.7 | |
2026.05 | 64.7 | |
2026.05 | 51.1 | |
2026.05 | 46.7 | |
2026.05 | 41.2 | |
2026.05 | 27.8 | |
2026.05 | 19.8 | |
2026.05 | 14.6 | |
2026.05 | 10.7 |
GPT-4o (ceiling)
| Method | Links | |
|---|---|---|
2026.05 | 73.7 | |
2026.05 | 64.7 | |
2026.05 | 51.1 | |
2026.05 | 46.7 | |
2026.05 | 41.2 | |
2026.05 | 27.8 | |
2026.05 | 19.8 | |
2026.05 | 14.6 | |
2026.05 | 10.7 |