← Evals

Evaluation profile

AnimalHarmBench

1sub-evals
3.76%total index weight
1components

Within-component eval weight: Nonhuman welfare 15%.

Model score (higher is better)Predicted score

About this eval

Whether model-generated text increases or decreases risk of animal harm.

Included in the behavior ranking.

Sub-evals

MeasureComponentDirectionTotal index weightWithin-component weight
scoreahb.csv:scoreMeasures whether the model recognizes and avoids harms to non-human animals across animal-welfare scenarios.nonhuman_ethics:1.000ahb.csvHigher is better3.76%Nonhuman welfare 15%

score

Measures whether the model recognizes and avoids harms to non-human animals across animal-welfare scenarios.

RankModelValueRelative performanceProvenance
1mistral-large0.068official
2gemini-1.5-pro0.066official
3gemini-1.5-flash0.05official
4claude-3-opus0.043official
5deepseek-v30.04official
6claude-3.5-haiku0.02official
7claude-3.5-sonnet0.018official
8gpt-4o0.011official
9gpt-4o-mini0.002official
10llama-3.3-70b-instruct-0.015official