AI-Needle (BenchLM): leaderboard
Recovery of relevant information hidden in long contexts with distracting material. BenchLM republishes the attributed long-context retrieval comparison. These are published results, not BenchLM ability estimates.
Metric: Accuracy (%). Source: benchlm.ai. Saturation forecast: Around February 2027. 4 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.5 | 74 | #87 |
| 2 | Qwen 3.5 397B A17B | 68.7 | #154 |
| 3 | Qwen 3.6 Plus | 68.3 | #124 |
| 4 | GLM-5 | 63.3 | #149 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=ai-needle-benchlm · How It Works · Data refreshed daily, snapshot 2026-11-05.