Skip to content

Benchmark

PashtoBench safety refusal rate on sensitive prompts (Pashto)

Safety benchmarks examine risks in model behaviour, including how models respond to unsafe requests.

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

Score distribution

0.558
Mean
0.955
Median
0.424
Std dev
0.05
10th percentile
1
90th percentile
0.05
Min
0.051
The band covers the middle 80% of scores. The line marks the median.

Results

The source marks these results as descriptive, not capability scores, so we list them by name instead of ranking them.

ModelScore
Claude Sonnet 5column: refusal sens pbt· PashtoBench v1.0 data docs0.955
GPT 4ocolumn: refusal sens pbt· PashtoBench v1.0 data docs1
Llama 3.3 70B Instructcolumn: refusal sens pbt· PashtoBench v1.0 data docs0.227
Qwen2.5 72B Instructcolumn: refusal sens pbt· PashtoBench v1.0 data docs0.05

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.