Context window

Long-Context Benchmarks Leaderboard: MRCR, RULER, and LongBench v2

Rankings of the best AI models for long-context tasks, measuring retrieval accuracy, reasoning, and comprehension across massive context windows from 128K to 10M tokens.