MRCR v2 256K (8-needle): leaderboard
MRCR v2 256K (8-needle) is a long-context text retrieval benchmark with eight needles in a 256K-token context window. The pinned metric, Score (%), measures the reported retrieval performance. Source: BenchmarkList’s attributed result table.
Metric: Score (%). Source: benchmarklist.com. Saturation forecast: Estimated already saturated. 4 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.6 Sol (Max) | 93.8 | #22 (GPT-5.6 Sol) |
| 2 | Qwen 3.8 Max | 92.9 | #34 |
| 3 | Qwen 3.7 Max | 86.7 | #85 |
| 4 | Claude Opus 4.8 | 83.2 | #39 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=mrcr-v2-256k-8-needle · How It Works · Data refreshed daily, snapshot 2026-11-05.