LongBenchPro

LongBenchPro - long context understanding benchmark with professional-level tasks.

Category

Long Context

Max Score

100.0

Score Type

percent

Active

Yes

Model Scores

Scores for LongBenchPro

Model Score Date Verified
MiniCPM5-2B
72.91%
— Verified
LFM2.5-2.6B
30.88%
— Verified
Qwen3.5-2B
8.20
— Verified
Gemma4-E2B
67.73%
— Verified
Qwen3.5-4B
100.00%
— Verified
granite-4.2-3B
52.99%
— Verified
Nemotron-3-Nano-4B
39.24%
— Verified
Gemma4-E4B
90.24%
— Verified
LFM2.5-8B-A1B
22.71%
— Verified