OpenAI MRCR v2 8-needle - 64K-128K
General QA
Eight-needle OpenAI retrieval averaged within the 64K-128K context band. Shares the family factor group; exact item count unreported.
Top models (higher is better)
| Model | Score |
|---|---|
| GPT-5.5 | 83.1 |
Benchmarks
218 sources · 1,060 models × 2,130 benchmarks · 109,661 scoresGeneral QA
Eight-needle OpenAI retrieval averaged within the 64K-128K context band. Shares the family factor group; exact item count unreported.
| Model | Score |
|---|---|
| GPT-5.5 | 83.1 |