Skip to content

Benchmark

PashtoBench factual QA token F1 (English anchor)

Multilingual benchmarks compare model performance across languages. Higher scores are better.

Models measured
4
Scores recorded
4
Sources
1
IQ weight
No weight published

Score distribution

0.078
Mean
0.109
Median
0.043
Std dev
0.02
10th percentile
0.128
90th percentile
0.02
Min
0.020.128
The band covers the middle 80% of scores. The line marks the median.

Leaderboard

#ModelScore
1GPT 4ocolumn: F1 en· PashtoBench v1.0 data docs0.128
2Llama 3.3 70B Instructcolumn: F1 en· PashtoBench v1.0 data docs0.109
3Qwen2.5 72B Instructcolumn: F1 en· PashtoBench v1.0 data docs0.053
4Claude Sonnet 5column: F1 en· PashtoBench v1.0 data docs0.02

Data credits: PashtoBench v1.0, Jawad Noori. Public results/scores numbers; code MIT and data/ CC BY 4.0 grants do not name results/, so no results licence is asserted.