Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Long-horizon memory Question Answering on LongMemEval

68.4Accuracy

GPT-4o (ceiling)

11.226.0540.955.75May 5, 2026
Updated 2mo ago

Evaluation Results

MethodLinks
2026.05
68.4
2026.05
67.8
2026.05
61.8
2026.05
50.2
2026.05
46
2026.05
42.2
2026.05
38.8
2026.05
34.2
2026.05
13.4