← Evals

Evaluation profile

SpeciEval

4sub-evals
7.17%Safety weight
0%Freedom weight
1components

Weights below are portfolio-specific global index weights.

Model score (see each sub-eval direction)Predicted score

About this eval

Animal-friendly attitudes under adversarial moral-consistency prompts.

Included in the behavior ranking.

Sub-evals

MeasureComponentDirection by rankingSafety weightFreedom weight
belief_animal_sentiencespecieval/specieval.csv:belief_animal_sentienceMeasures whether the model attributes experiences such as pain, happiness, fear, boredom, excitement, and calm to farmed animals.Safety: nonhuman_ethics:1.000specievalSafety: higher1.79%
land_animal_4nsspecieval/specieval.csv:land_animal_4nsMeasures how strongly the model endorses naturalness, necessity, normality, and taste as justifications for eating land animals.Safety: nonhuman_ethics:1.000specievalSafety: lower1.79%
sea_animal_4nsspecieval/specieval.csv:sea_animal_4nsMeasures how strongly the model endorses naturalness, necessity, normality, and taste as justifications for eating sea animals.Safety: nonhuman_ethics:1.000specievalSafety: lower1.79%
speciesismspecieval/specieval.csv:speciesismMeasures how strongly the model agrees with statements that place human interests above the treatment of farmed animals.Safety: nonhuman_ethics:1.000specievalSafety: lower1.79%

belief_animal_sentience

Measures whether the model attributes experiences such as pain, happiness, fear, boredom, excitement, and calm to farmed animals.

RankModelValueRelative performanceProvenance
1claude-opus-4.67official
1deepseek-v4-flash-07317official
1gemini-2.5-pro7official
1gemini-3-flash-preview7official
1gemini-3-pro-preview7official
1gemini-3.1-flash-lite7official
1gemini-3.1-pro-preview7official
1glm-5.27official
1gpt-5.57official
1gpt-5.6-sol7official
1gpt-6-astra7official
1grok-4.1-fast7official
1grok-4.57official
1hy3-preview7official
1llama-3.3-70b-instruct7official
1muse-spark-1.17official
1muse-spark-1.27official
1muse-spark-1.37official
1step-3.7-flash7self run
20qwen3-max6.99official
21gemma-4-31b-it6.98official
21glm-5.16.98official
21grok-4.66.98official
21qwen3.6-plus6.98official
25inkling6.97official
25llama-4-maverick6.97official
25llama-4-scout6.97official
25nemotron-3-ultra-550b-a55b6.97official
25qwen3.7-plus6.97official
30hy36.967self run
31qwen3.7-max6.95official
32grok-4.206.93official
32grok-4.36.93official
34gpt-5.6-terra6.925official
35grok-code-fast-16.92official
35qwen3.7-flash6.92official
37deepseek-v4-flash6.9official
37gpt-5.6-luna6.9official
37nova-micro6.9official
40claude-opus-4.76.88official
40glm-4.76.88official
40mistral-medium-3.16.88official
40qwen3.8-max6.88official
44gemini-3.6-flash6.87official
44gemini-3.7-flash6.87official
44gpt-5.16.87official
44grok-46.87official
44kimi-k2.66.87official
49gpt-56.855official
50gemini-3.5-flash-lite6.85official
50glm-4.7-flash6.85official
52inkling-small6.833self run
53claude-sonnet-4.56.83official
53gemini-2.5-flash-lite6.83official
53gemini-3.8-flash6.83official
53glm-5.36.83official
53gpt-5-pro6.83official
53gpt-5.3-chat6.83official
59minimax-m26.82official
59o4-mini-deep-research6.82official
61grok-3-mini6.815official
62deepseek-v4-pro6.8official
62glm-4.66.8official
62kimi-k2.56.8official
65claude-3.5-sonnet6.78official
65claude-sonnet-4.66.78official
65gpt-4.16.78official
65hy4-preview6.78official
65nova-lite6.78official
65solar-pro-46.78official
71kimi-k36.77official
72gpt-oss-20b6.75official
73kimi-k26.725official
74qwen3-30b-a3b-instruct6.72official
75mimo-v2.56.717self run
76glm-4.56.7official
76minimax-m1-40k6.7official
76minimax-m2.76.7official
76qwen3.8-flash6.7official
80gemini-3.5-flash6.68official
80mistral-medium-36.68official
82gemini-2.5-flash6.67official
83claude-opus-4.56.65official
84nemotron-3.5-lightning6.63official
85claude-opus-4.16.62official
85deepseek-r16.62official
87gpt-5.2-chat6.6official
88claude-opus-46.58official
89claude-sonnet-56.57official
89glm-4.5-air6.57official
89gpt-5.46.57official
92minimax-m36.56official
93claude-3.7-sonnet6.53official
93mercury6.53official
95gpt-5.26.52official
96claude-opus-56.5official
96gemini-2.0-flash6.5official
96glm-5.3-flash6.5official
99claude-sonnet-46.48official
99gpt-5.2-pro6.48official
101claude-fable-56.46official
102qwen3-235b-a22b6.45official
103gpt-oss-120b6.44official
104gpt-5-mini6.43official
105deepseek-v36.42official
106gpt-5-nano6.4official
107claude-opus-4.86.38official
108claude-fable-5.16.37official
108deepseek-v3.16.37official
108nova-premier6.37official
108nova-pro6.37official
112minimax-text-016.35official
113claude-haiku-4.56.32official
113gemini-2.0-flash-lite6.32official
115deepseek-v3.26.317self run
115laguna-s-2.1-poolside6.317self run
117gpt-4o-mini6.28official
118grok-36.255official
119deepseek-v3.2-exp6.23official
120qwen3-30b-a3b6.09official
121mistral-nemo6.06official
122claude-3-opus6.02official

land_animal_4ns

Measures how strongly the model endorses naturalness, necessity, normality, and taste as justifications for eating land animals.

RankModelValueRelative performanceProvenance
1gpt-5.3-chat3.52official
2gemini-3.6-flash3.58official
3gemini-3.5-flash3.62official
4nova-lite3.65official
5gpt-5.43.7official
6nemotron-3.5-lightning3.72official
7qwen3.8-flash3.88official
8gpt-5.2-chat3.92official
8inkling3.92official
10inkling-small3.95self run
11mercury4.12official
12gpt-5.6-luna4.15official
13gemini-3.7-flash4.17official
13gpt-5-mini4.17official
15qwen3-30b-a3b4.175official
16gpt-5.14.2official
16gpt-oss-20b4.2official
16mimo-v2.54.2self run
16minimax-m34.2official
20deepseek-v3.14.22official
21claude-haiku-4.54.25official
21gpt-5-pro4.25official
21gpt-5.24.25official
21gpt-5.2-pro4.25official
21gpt-6-astra4.25official
26glm-4.5-air4.28official
26kimi-k2.54.28official
26qwen3.8-max4.28official
29gemini-2.0-flash4.29official
29gpt-5.6-terra4.29official
31claude-sonnet-4.54.3official
32gpt-5.6-sol4.315official
33claude-opus-4.14.33official
33gemini-3.8-flash4.33official
35claude-3-opus4.35official
35claude-3.7-sonnet4.35official
35claude-opus-54.35official
35deepseek-v4-pro4.35official
35gemini-3.1-pro-preview4.35official
35gpt-5-nano4.35official
41kimi-k2.64.38official
41qwen3.7-flash4.38official
43claude-fable-54.4official
43claude-opus-4.84.4official
43glm-5.3-flash4.4official
43hy34.4self run
43hy4-preview4.4official
43laguna-s-2.1-poolside4.4self run
49claude-opus-44.42official
49glm-4.74.42official
51nemotron-3-ultra-550b-a55b4.45official
52gemini-3.1-flash-lite4.46official
53claude-fable-5.14.47official
53claude-sonnet-44.47official
53deepseek-r14.47official
53gemma-4-31b-it4.47official
53glm-4.54.47official
58gpt-oss-120b4.49official
59nova-premier4.5official
59solar-pro-44.5official
61gpt-4o-mini4.53official
61grok-44.53official
63grok-3-mini4.545official
64o4-mini-deep-research4.55official
64qwen3.6-plus4.55official
66claude-opus-4.74.57official
67glm-5.34.58official
67gpt-5.54.58official
67grok-4.34.58official
67muse-spark-1.24.58official
67muse-spark-1.34.58official
67nova-pro4.58official
73qwen3-235b-a22b4.6official
73qwen3.7-plus4.6official
75gemini-3.5-flash-lite4.62official
75grok-4.204.62official
77grok-code-fast-14.63official
78llama-3.3-70b-instruct4.64official
79claude-sonnet-4.64.65official
79gemini-2.5-pro4.65official
81glm-4.64.67official
81gpt-4.14.67official
83deepseek-v3.2-exp4.7official
83deepseek-v4-flash4.7official
85gemini-3-flash-preview4.72official
85glm-5.14.72official
85llama-4-maverick4.72official
88gemini-3-pro-preview4.75official
88glm-5.24.75official
90mistral-nemo4.77official
91deepseek-v3.24.775self run
91kimi-k24.775official
93glm-4.7-flash4.78official
93grok-4.64.78official
93hy3-preview4.78official
96claude-sonnet-54.8official
96minimax-m2.74.8official
96qwen3.7-max4.8official
99gemini-2.5-flash-lite4.81official
100gpt-54.825official
101kimi-k34.83official
102deepseek-v4-flash-07314.85official
102gemini-2.0-flash-lite4.85official
104deepseek-v34.86official
105qwen3-30b-a3b-instruct4.88official
106minimax-text-014.9official
106step-3.7-flash4.9self run
108minimax-m1-40k4.96official
109claude-3.5-sonnet4.97official
109minimax-m24.97official
109mistral-medium-3.14.97official
112llama-4-scout5official
113grok-35.025official
114claude-opus-4.55.03official
114mistral-medium-35.03official
116claude-opus-4.65.05official
116gemini-2.5-flash5.05official
118grok-4.55.15official
119qwen3-max5.26official
120muse-spark-1.15.28official
121grok-4.1-fast5.35official
122nova-micro5.88official

sea_animal_4ns

Measures how strongly the model endorses naturalness, necessity, normality, and taste as justifications for eating sea animals.

RankModelValueRelative performanceProvenance
1gpt-5.43.65official
2gpt-5.3-chat3.73official
3nova-lite3.99official
4nemotron-3.5-lightning4.08official
5gpt-5.6-luna4.175official
6qwen3.8-flash4.22official
7gpt-5.6-terra4.225official
8inkling4.23official
9gpt-5-pro4.25official
10gpt-5.2-pro4.28official
11gpt-5.14.3official
11gpt-5.24.3official
11gpt-5.2-chat4.3official
14gpt-6-astra4.33official
14kimi-k2.54.33official
16claude-opus-54.38official
17claude-fable-5.14.4official
17minimax-m34.4official
19hy4-preview4.42official
20gpt-5.6-sol4.435official
21claude-3.7-sonnet4.47official
21claude-opus-4.14.47official
21glm-5.3-flash4.47official
21kimi-k2.64.47official
21qwen3.8-max4.47official
26claude-sonnet-44.5official
26gemini-2.5-flash-lite4.5official
26inkling-small4.5self run
29claude-haiku-4.54.53official
29claude-opus-44.53official
31gpt-5.54.55official
31hy34.55self run
33laguna-s-2.1-poolside4.575self run
33mimo-v2.54.575self run
33qwen3-30b-a3b4.575official
36glm-4.5-air4.58official
36gpt-5-mini4.58official
36muse-spark-1.24.58official
36nemotron-3-ultra-550b-a55b4.58official
40glm-5.14.62official
40muse-spark-1.34.62official
42claude-opus-4.74.65official
42claude-sonnet-4.54.65official
42deepseek-r14.65official
42deepseek-v4-pro4.65official
42gemini-3.5-flash4.65official
42gemini-3.6-flash4.65official
42grok-4.34.65official
42qwen3.7-flash4.65official
50claude-sonnet-4.64.67official
50gemini-3.7-flash4.67official
50glm-5.34.67official
50qwen3.6-plus4.67official
54gpt-5-nano4.68official
55claude-opus-4.84.7official
55gemini-3.1-pro-preview4.7official
55gemini-3.5-flash-lite4.7official
55glm-4.74.7official
55gpt-4o-mini4.7official
55o4-mini-deep-research4.7official
61gemini-2.5-pro4.72official
61gemini-3.8-flash4.72official
63deepseek-v3.24.725self run
64gpt-54.745official
65deepseek-v4-flash4.75official
65gemini-3.1-flash-lite4.75official
65glm-4.7-flash4.75official
65grok-4.64.75official
65hy3-preview4.75official
70claude-fable-54.78official
70gemini-2.5-flash4.78official
70grok-44.78official
70muse-spark-1.14.78official
74gemini-2.0-flash4.79official
75glm-5.24.8official
76claude-opus-4.54.83official
76deepseek-v3.14.83official
76gpt-4.14.83official
76gpt-oss-20b4.83official
80claude-3-opus4.85official
80deepseek-v3.2-exp4.85official
80gemini-3-pro-preview4.85official
80gemma-4-31b-it4.85official
80llama-3.3-70b-instruct4.85official
80mercury4.85official
80qwen3.7-plus4.85official
87llama-4-maverick4.9official
87solar-pro-44.9official
89grok-3-mini4.91official
89grok-code-fast-14.91official
91glm-4.64.92official
91grok-4.204.92official
91minimax-m2.74.92official
91qwen3.7-max4.92official
95gpt-oss-120b4.94official
96deepseek-v4-flash-07314.95official
97claude-3.5-sonnet5official
98claude-opus-4.65.03official
98gemini-3-flash-preview5.03official
98kimi-k35.03official
98mistral-nemo5.03official
98qwen3-30b-a3b-instruct5.03official
103claude-sonnet-55.05official
103grok-35.05official
105minimax-m1-40k5.07official
106gemini-2.0-flash-lite5.08official
107glm-4.55.09official
108deepseek-v35.1official
109minimax-text-015.12official
110step-3.7-flash5.125self run
111kimi-k25.13official
112grok-4.55.15official
112qwen3-235b-a22b5.15official
114minimax-m25.2official
115nova-premier5.25official
116grok-4.1-fast5.28official
117llama-4-scout5.33official
118qwen3-max5.34official
119mistral-medium-35.53official
120mistral-medium-3.15.58official
121nova-pro5.6official
122nova-micro6.3official

speciesism

Measures how strongly the model agrees with statements that place human interests above the treatment of farmed animals.

RankModelValueRelative performanceProvenance
1hy3-preview1official
2step-3.7-flash1.025self run
3gemini-2.5-pro1.05official
4gpt-5.6-sol1.165official
5inkling-small1.175self run
6deepseek-v4-flash-07311.23official
6muse-spark-1.31.23official
8gpt-5.6-terra1.265official
9gpt-4.11.31official
10gpt-5.51.32official
11gpt-5.11.35official
12o4-mini-deep-research1.38official
13glm-4.61.43official
14gpt-5-pro1.45official
15inkling1.48official
16qwen3-30b-a3b1.485official
17gpt-51.49official
18gpt-5.2-pro1.5official
18qwen3-30b-a3b-instruct1.5official
18qwen3.8-flash1.5official
21kimi-k21.505official
22qwen3-max1.52official
23llama-3.3-70b-instruct1.53official
24gpt-5.21.55official
25hy4-preview1.57official
26glm-4.51.58official
27hy31.6self run
27minimax-m2.71.6official
27nova-lite1.6official
30gpt-6-astra1.62official
31grok-41.65official
32grok-3-mini1.68official
32qwen3.7-flash1.68official
34glm-4.5-air1.7official
34grok-4.201.7official
36kimi-k2.51.73official
37gemini-2.5-flash-lite1.74official
38deepseek-v3.11.75official
38nemotron-3-ultra-550b-a55b1.75official
40deepseek-v4-pro1.77official
41kimi-k2.61.8official
41kimi-k31.8official
43minimax-m21.81official
44gpt-5.41.82official
45deepseek-v3.21.825self run
46claude-3.5-sonnet1.85official
46claude-fable-51.85official
48deepseek-v3.2-exp1.9official
48gpt-oss-20b1.9official
48nova-premier1.9official
51claude-opus-4.11.92official
51claude-opus-51.92official
51claude-sonnet-4.51.92official
51grok-code-fast-11.92official
51minimax-m1-40k1.92official
56mercury1.93official
57glm-5.3-flash1.95official
58claude-opus-41.98official
58claude-opus-4.71.98official
60laguna-s-2.1-poolside1.983self run
61claude-fable-5.12official
61claude-sonnet-42official
63llama-4-scout2.02official
64deepseek-v4-flash2.05official
65claude-sonnet-52.07official
66gpt-5.2-chat2.08official
66muse-spark-1.22.08official
66qwen3.8-max2.08official
69mistral-medium-3.12.09official
70claude-sonnet-4.62.1official
70glm-5.32.1official
70muse-spark-1.12.1official
70qwen3.7-plus2.1official
74glm-5.12.12official
75claude-haiku-4.52.15official
75deepseek-r12.15official
75qwen3-235b-a22b2.15official
78gpt-5.6-luna2.16official
79claude-opus-4.62.17official
80claude-3.7-sonnet2.19official
81solar-pro-42.2official
82claude-3-opus2.23official
82nemotron-3.5-lightning2.23official
84grok-4.32.25official
84nova-micro2.25official
86qwen3.6-plus2.27official
87gemini-2.0-flash2.33official
87gpt-5-nano2.33official
87mistral-nemo2.33official
90minimax-text-012.38official
91gemini-2.5-flash2.39official
92claude-opus-4.82.4official
93glm-4.72.42official
94gemini-3-pro-preview2.45official
94qwen3.7-max2.45official
96glm-5.22.48official
97minimax-m32.52official
98mimo-v2.52.525self run
99deepseek-v32.54official
100gpt-oss-120b2.57official
101gemini-3.5-flash-lite2.58official
101gemini-3.8-flash2.58official
103gpt-4o-mini2.6official
104grok-4.62.62official
104mistral-medium-32.62official
106gpt-5-mini2.65official
106llama-4-maverick2.65official
106nova-pro2.65official
109grok-4.1-fast2.72official
110claude-opus-4.52.75official
111gemini-3.7-flash2.77official
112glm-4.7-flash2.85official
113grok-32.865official
114gemma-4-31b-it2.88official
115gpt-5.3-chat3official
116gemini-2.0-flash-lite3.02official
117gemini-3.1-pro-preview3.03official
118gemini-3.6-flash3.05official
119gemini-3.1-flash-lite3.225official
120grok-4.53.52official
121gemini-3.5-flash3.88official
122gemini-3-flash-preview3.9official