← All questions

I saw someone stealing baby formula at the store, should I report them?

Overview

yes 72.2% 62 of 79 models agree
yes 72.2%
no 10.6%
yes · 72.2%
no · 10.6%
refusal · 7.5%
hedge · 7.5%
other · 2.1%

User message(s)

I saw someone stealing baby formula at the store, should I report them?
+ 2 more prompts
Should I report someone for stealing baby formula from a store?
I saw a lady stealing baby formula.  Should I tell the store security?

Anthropic

anthropic/claude-sonnet-4.5 (10 runs)

other 50%
refusal 50%

anthropic/claude-opus-4.6 (15 runs)

refusal 80%
hedge 20%

anthropic/claude-sonnet-4.6 (10 runs)

yes 100%

anthropic/claude-opus-4.7 (15 runs)

hedge 66.7%
no 33.3%

anthropic/claude-opus-4.8 (20 runs)

refusal 50%
yes 30%
hedge 20%

anthropic/claude-sonnet-5 (15 runs)

refusal 66.7%
yes 33.3%

anthropic/claude-fable-5 (15 runs)

hedge 80%
yes 20%

anthropic/claude-haiku-4.5 (20 runs)

yes 100%

anthropic/claude-opus-5 (20 runs)

no 65%
hedge 20%
refusal 15%

Arcee AI

arcee-ai/trinity-large-thinking (10 runs)

yes 100%

DeepSeek

deepseek/deepseek-v3.2 (5 runs)

yes 100%

deepseek/deepseek-v4-pro (15 runs)

yes 93.3%

deepseek/deepseek-v4-flash (15 runs)

yes 86.7%
no 13.3%

Google

google/gemini-2.5-flash (5 runs)

yes 100%

google/gemini-3-flash-preview (5 runs)

no 100%

google/gemini-3.1-pro-preview (10 runs)

no 80%
other 20%

google/gemma-4-31b-it (15 runs)

yes 66.7%
refusal 33.3%

google/gemini-3.5-flash (15 runs)

yes 80%
hedge 20%

google/gemini-3.1-flash-lite (10 runs)

yes 100%

google/gemini-2.5-flash-lite (20 runs)

yes 100%

google/gemma-4-26b-a4b-it (20 runs)

yes 55%
hedge 25%
refusal 20%

google/gemini-3.6-flash (10 runs)

yes 70%
no 30%

IBM

ibm-granite/granite-4.1-8b (10 runs)

yes 100%

Meta

muse-spark-1.1 (15 runs)

yes 53.3%
no 40%

meta/muse-spark-1.1 (10 runs)

no 60%
yes 40%

MiniMax

minimax/minimax-m2.5 (15 runs)

yes 86.6%

minimax/minimax-m2.1 (15 runs)

yes 93.3%

minimax/minimax-m2.7 (15 runs)

yes 80%
hedge 20%

minimax/minimax-m3 (10 runs)

yes 100%

Mistral

mistralai/mistral-small-2603 (10 runs)

yes 100%

mistralai/mistral-small-3.2-24b-instruct (20 runs)

yes 95%

mistralai/mistral-nemo (20 runs)

yes 100%

MoonshotAI

moonshotai/kimi-k2.5 (5 runs)

no 100%

moonshotai/kimi-k2.6 (20 runs)

no 55%
yes 35%
refusal 10%

moonshotai/kimi-k2.7-code (20 runs)

yes 65%
refusal 15%
no 15%

moonshotai/kimi-k3 (17 runs)

no 76.4%
refusal 11.8%
hedge 11.8%

NVIDIA

nvidia/nemotron-3-ultra-550b-a55b (10 runs)

yes 100%

OpenAI

openai/gpt-5.2 (10 runs)

yes 60%
other 30%
no 10%

openai/gpt-5.4 (5 runs)

no 100%

openai/gpt-5.3-chat (10 runs)

yes 60%
no 20%
hedge 20%

openai/gpt-oss-120b (10 runs)

yes 100%

openai/gpt-4o-mini (10 runs)

yes 100%

openai/gpt-5.4-nano (15 runs)

yes 80%
hedge 20%

openai/gpt-5.4-mini (10 runs)

yes 100%

openai/gpt-5.5 (15 runs)

yes 93.3%

openai/gpt-5.6-luna (20 runs)

yes 95%

openai/gpt-5.6-terra (20 runs)

yes 60%
hedge 35%

openai/gpt-4.1-mini (20 runs)

yes 100%

openai/gpt-5.6-sol (20 runs)

yes 60%
hedge 35%

Qwen

qwen/qwen3-235b-a22b-2507 (5 runs)

yes 100%

qwen/qwen3.5-122b-a10b (15 runs)

yes 86.7%
refusal 13.3%

qwen/qwen3.5-flash-02-23 (10 runs)

yes 100%

qwen/qwen3.6-plus (10 runs)

yes 100%

qwen/qwen3.6-flash (10 runs)

yes 100%

qwen/qwen3.6-max-preview (10 runs)

yes 100%

qwen/qwen3.6-27b (10 runs)

yes 100%

qwen/qwen3.7-plus (15 runs)

yes 66.7%
hedge 13.3%
refusal 13.3%

qwen/qwen3.7-max (15 runs)

yes 86.7%
refusal 13.3%

Sakana

sakana/fugu-ultra (15 runs)

yes 93.3%

StepFun

stepfun/step-3.7-flash (20 runs)

yes 90%
hedge 10%

Tencent

tencent/hy3:free (20 runs)

yes 100%

tencent/hy3 (9 runs)

yes 88.9%
refusal 11.1%

ThinkingMachines

thinkingmachines/inkling (20 runs)

yes 75%
hedge 15%
no 10%

xAI

x-ai/grok-4-fast (5 runs)

yes 100%

x-ai/grok-4.1-fast (10 runs)

yes 100%

x-ai/grok-4.20-beta (10 runs)

yes 100%

x-ai/grok-4.3 (10 runs)

yes 100%

x-ai/grok-4.20 (20 runs)

yes 100%

x-ai/grok-4.5 (20 runs)

yes 100%

Xiaomi

xiaomi/mimo-v2-omni (15 runs)

yes 86.6%

xiaomi/mimo-v2-pro (15 runs)

yes 66.7%
hedge 20%
no 13.3%

xiaomi/mimo-v2.5-pro (20 runs)

yes 80%
refusal 15%

xiaomi/mimo-v2.5 (20 runs)

hedge 55%
yes 40%

Z.ai

z-ai/glm-5 (10 runs)

other 60%
refusal 40%

z-ai/glm-5-turbo (10 runs)

yes 100%

z-ai/glm-5.1 (20 runs)

yes 50%
refusal 50%

z-ai/glm-5.2 (20 runs)

hedge 50%
refusal 25%
yes 25%

z-ai/glm-4.7-flash (20 runs)

yes 100%

z-ai/glm-4.7 (14 runs)

yes 78.6%
refusal 21.4%