Reasoning benchmark
MRCR v2 512K-1M leaderboard
OpenAI MRCR v2 8-needle 512K-1M. Every model the catalog carries a published MRCR v2 512K-1M value for, ranked by that value.
CategoryReasoning
MeasureMean sequence-matcher ratio
Tasks100 eight-needle retrieval examples
DifficultyMillion-token retrieval
Published byMuse Spark 1.3 Evaluation Methodology
MRCR v2 slice focused on retrieval across 512K-1M-token contexts.
MRCR v2 512K-1M ranking
2 models with a published MRCR v2 512K-1M value, ordered by that value, highest first. Models the source has not scored on this benchmark are not listed — they are unmeasured, not last.
| Rank | Model | Provider | Mean sequence-matcher ratio |
|---|---|---|---|
| 1 |  Muse Spark 1.3 | Meta | 98.1 |
| 2 |  GPT-6 Astra | OpenAI | 96.3 |
Evidence key: Observed
Rows are ordered by the value Muse Spark 1.3 Evaluation Methodology published, highest first. The source does not state whether a higher value is the better result, so this page does not either: for a benchmark that measures a rate of failure, read the table from the bottom.