Beta version: *Information might not be fully accurate. Please report any discrepancies.

LongBench v2

Name: LongBench v2 leaderboard dataset
Creator: LLM Registry

longbench-v2Long ContextHigher is better

Comprehensive long-context understanding (128k).

Loading chart…

Loading leaderboard…

Benchmark Info

Current SOTA

Median Score40.8%

DistributionLimited data

Score Distribution

P10

P50

P90

0100%

MRCR v2

mrcr-v2

AA-LCR

lcr

Graphwalks Bfs

graphwalks-bfs