BABILong (NIAH, v0, 128k): leaderboard
Metric: QA1–5 Mean Accuracy (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 16 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Llama 3.1 70B Instruct | 45 |
| 2 | Llama 3.1 8B Instruct | 39 |
| 3 | c4ai-command-r-v01 | 38 |
| 4 | GPT-4 Preview (0125) | 36 |
| 5 | Phi-3-medium-128k-instruct | 30 |
| 6 | Yi-9B-200K | 24 |
| 7 | Yarn-Mistral-7B-128k | 9 |
| 8 | Phi-3 Mini 128K Instruct | 7 |
Interactive version: theaggregate.ai/benchmark?slug=babilong-niah-v0-128k · How It Works · Data refreshed daily, snapshot 2026-10-09.