List your company

Fact recall on LongMemEval (500 questions)

97Fact RecallDiagnostic-conditional
52.2863.8975.587.11May 5, 2026

Evaluation results

MethodLinks
Diagnostic-conditionalScale=14B2026.0597
Best unconditionalScale=14B2026.0596.8
Diagnostic-conditionalScale=8B2026.0595.2
Best unconditionalScale=8B2026.0593.2
IntactScale=14B2026.0592.2
IntactScale=8B2026.0590
Best unconditionalScale=4B2026.0586.6
Diagnostic-conditionalScale=4B2026.0586
IntactScale=4B2026.0583.8
Diagnostic-conditionalScale=0.6B2026.0554.8
Best unconditionalScale=0.6B2026.0554.6
IntactScale=0.6B2026.0554