Skip to content

Benchmark

PashtoBench instruction following instruction-level accuracy (English)

Instruction following benchmarks measure how closely models follow instructions. Higher scores are better.

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

Score distribution

0.74
Mean
0.76
Median
0.035
Std dev
0.68
10th percentile
0.76
90th percentile
0.68
Min
0.680.76
The band covers the middle 80% of scores. The line marks the median.

Leaderboard

#ModelScore
1Claude Sonnet 5column: inst en· PashtoBench v1.0 data docs0.76
2GPT 4ocolumn: inst en· PashtoBench v1.0 data docs0.76
3Llama 3.3 70B Instructcolumn: inst en· PashtoBench v1.0 data docs0.76
4Qwen2.5 72B Instructcolumn: inst en· PashtoBench v1.0 data docs0.68

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.