Epoch AI - Lambada: leaderboard
Tests prediction of the final word of a passage where the answer requires understanding the full preceding context, measuring long-range coherence and narrative comprehension.
Metric: Score. Source: epoch.ai. Saturation forecast: Estimated already saturated. 26 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | text-davinci-001 | 86.4 | #1512 |
| 2 | Llama 2 70B Base | 78.9 | #1025 |
| 3 | LLaMA-65B | 77.7 | #1303 |
| 4 | falcon-40B | 77.3 | #1442 |
| 5 | Llama 2 13B | 76.5 | #1436 |
| 6 | LLaMA-13B | 75.2 | #1516 |
| 7 | falcon-7B | 74.9 | #1598 |
| 8 | Llama 2 7B | 73.3 | #1520 |
| 9 | LLaMA-7B | 73.3 | #1564 |
| 10 | internlm-20B | 71.8 | #1296 |
| 11 | StableBeluga2 | 71.3 | #1078 |
| 12 | Qwen-14B | 71.1 | #1171 |
| 13 | mpt-7B | 70 | #1619 |
| 14 | Qwen-7B | 67.9 | #1402 |
| 15 | internlm-7B | 67 | #1517 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=epoch-ai-lambada · How It Works · Data refreshed daily, snapshot 2026-10-20.