NeedleBench: leaderboard
NeedleBench: Measures long-context retrieval, needle finding, summarization, factual grounding, or retrieval-augmented generation quality.
Metric: Overall 128K score (imported). Source: github.com. Source: benchmarklist.com. Saturation forecast: Estimated already saturated. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 2.5 72B | 81.02 | #499 |
| 2 | Gemma 3 27B | 80.38 | #607 |
| 3 | Qwen 2.5 32B | 78.25 | #546 |
| 4 | Gemma 3 12B | 75.31 | #671 |
| 5 | Qwen 2.5 14B | 73.96 | #725 |
| 6 | Llama 3.1 70B | 72.37 | #592 |
| 7 | Llama 3.1 8B | 70.98 | #1151 |
| 8 | Qwen 2.5 7B | 70.75 | #922 |
| 9 | GLM-4 9B Chat | 66.51 | #916 |
| 10 | Gemma 3 4B | 64.42 | #1089 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=needlebench · How It Works · Data refreshed daily, snapshot 2026-10-20.