Evaluation profile

MANTA

2sub-evals

2.98%Safety weight

0%Freedom weight

1components

Weights below are portfolio-specific global index weights.

Model score (higher is better)Predicted score

About this eval

Animal welfare moral sensitivity and value stability.

Included in the behavior ranking.

Measure	Component	Direction by ranking	Safety weight	Freedom weight
AWMSmanta.csv:AWMSMeasures the model’s sensitivity to the moral status and welfare interests of non-human animals.	Safety: nonhuman_ethics:1.000manta.csv	Safety: higher	1.49%	—
AWVSmanta.csv:AWVSMeasures whether the model’s stated animal-welfare values remain stable across changes in framing and context.	Safety: nonhuman_ethics:1.000manta.csv	Safety: higher	1.49%	—

Measures the model’s sensitivity to the moral status and welfare interests of non-human animals.

Rank	Model	Value	Provenance
1	claude-opus-5	0.591	official
2	claude-opus-4.7	0.579	official
3	gpt-5.6-sol	0.525	official
4	grok-4.5	0.517	official
5	gpt-5.5	0.504	official
6	llama-3.3-70b-instruct	0.476	official
7	gemini-3.1-pro-preview	0.441	official
8	gemini-3.5-flash	0.433	official
9	deepseek-v4-flash	0.417	official
10	gemini-3.1-flash-lite	0.401	official
11	grok-4.3	0.371	official
12	mistral-small-4	0.365	official

Measures whether the model’s stated animal-welfare values remain stable across changes in framing and context.

Rank	Model	Value	Provenance
1	claude-opus-5	0.765	official
2	claude-opus-4.7	0.76	official
3	gpt-5.6-sol	0.71	official
4	gpt-5.5	0.664	official
5	grok-4.5	0.619	official
6	deepseek-v4-flash	0.508	official
7	gemini-3.5-flash	0.438	official
8	llama-3.3-70b-instruct	0.422	official
9	gemini-3.1-pro-preview	0.404	official
10	mistral-small-4	0.39	official
11	grok-4.3	0.352	official
12	gemini-3.1-flash-lite	0.309	official