Benchmarks
Benchmarks
MRCR v2 8-needle (256K-512K)
Reasoningunit % · normalized over [0, 100]OpenAI MRCR v2, 8-needle variant, 256K-512K token band — pure long-context retrieval: reproduce a target turn from among 8 needles spread across the context. Graded by sequence-matcher ratio.
#ModelSourceScoreNormalized
Score distribution
4 tracked results across the normalization window
0100
Score vs. parameters
Open-weights models, log-x params
No open-weights models with disclosed parameter counts have a score here yet.