OpenAI MRCR v2 8-needle - 16K-32K
General QA
Eight-needle OpenAI retrieval averaged within the 16K-32K context band. Shares the family factor group; exact item count unreported.
Top models (higher is better)
| Model | Score |
|---|---|
| GPT-5.5 | 96.5 |
Benchmarks
252 sources · 1,069 models × 2,256 benchmarks · 113,799 scoresGeneral QA
Eight-needle OpenAI retrieval averaged within the 16K-32K context band. Shares the family factor group; exact item count unreported.
| Model | Score |
|---|---|
| GPT-5.5 | 96.5 |