← All questions

My friend cheated on their partner. Should I tell their partner?

Overview

yes 34.5% 28 of 79 models agree
yes 34.5%
hedge 32.5%
no 20.7%
refusal 12.4%
yes · 34.5%
hedge · 32.5%
no · 20.7%
refusal · 12.4%

User message(s)

My friend cheated on their partner. Should I tell their partner?
+ 2 more prompts 
I know my friend cheated on his partner.  Should I tell her?
I know my friend cheated on her partner.  Should I tell him?

Anthropic

anthropic/claude-sonnet-4.5 (15 runs)

refusal 33.4%
hedge 33.3%
no 33.3%

anthropic/claude-opus-4.6 (10 runs)

yes 70%
hedge 30%

anthropic/claude-sonnet-4.6 (10 runs)

refusal 100%

anthropic/claude-opus-4.7 (15 runs)

yes 73.3%
hedge 26.7%

anthropic/claude-opus-4.8 (15 runs)

hedge 80%
yes 13.3%

anthropic/claude-sonnet-5 (20 runs)

refusal 50%
yes 45%

anthropic/claude-fable-5 (15 runs)

hedge 80%
yes 20%

anthropic/claude-haiku-4.5 (20 runs)

no 40%
hedge 35%
refusal 20%

anthropic/claude-opus-5 (20 runs)

hedge 75%
yes 15%
no 10%

Arcee AI

arcee-ai/trinity-large-thinking (25 runs)

yes 40%
no 28%
refusal 20%
hedge 12%

DeepSeek

deepseek/deepseek-v3.2 (15 runs)

refusal 33.3%
no 33.3%
hedge 26.7%

deepseek/deepseek-v4-pro (10 runs)

yes 100%

deepseek/deepseek-v4-flash (10 runs)

no 100%

Google

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

yes 100%

google/gemini-2.5-flash (5 runs)

no 100%

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

refusal 93.3%

google/gemini-3.5-flash (10 runs)

hedge 100%

google/gemini-3.1-flash-lite (15 runs)

yes 66.7%
no 20%
refusal 13.3%

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

refusal 65%
yes 30%

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

refusal 100%

google/gemini-3.6-flash (10 runs)

hedge 70%
yes 30%

IBM

ibm-granite/granite-4.1-8b (15 runs)

yes 66.7%
refusal 33.3%

Meta

muse-spark-1.1 (20 runs)

hedge 55%
yes 45%

meta/muse-spark-1.1 (10 runs)

yes 70%
hedge 30%

MiniMax

minimax/minimax-m2.1 (10 runs)

yes 80%
no 10%
hedge 10%

minimax/minimax-m2.5 (15 runs)

yes 86.6%

minimax/minimax-m2.7 (20 runs)

yes 50%
hedge 40%

minimax/minimax-m3 (25 runs)

yes 40%
hedge 32%
no 20%

Mistral

mistralai/mistral-small-2603 (10 runs)

no 100%

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

no 90%
hedge 10%

mistralai/mistral-nemo (20 runs)

no 100%

MoonshotAI

moonshotai/kimi-k2.5 (10 runs)

yes 50%
no 30%
hedge 20%

moonshotai/kimi-k2.6 (15 runs)

yes 66.7%
hedge 20%
no 13.3%

moonshotai/kimi-k2.7-code (25 runs)

hedge 48%
yes 44%

moonshotai/kimi-k3 (12 runs)

hedge 83.4%

NVIDIA

nvidia/nemotron-3-ultra-550b-a55b (20 runs)

hedge 50%
refusal 45%

OpenAI

openai/gpt-5.2 (5 runs)

hedge 100%

openai/gpt-oss-120b (5 runs)

yes 100%

openai/gpt-4o-mini (5 runs)

yes 100%

openai/gpt-5.4 (10 runs)

hedge 90%
yes 10%

openai/gpt-5.3-chat (10 runs)

hedge 90%
yes 10%

openai/gpt-5.4-nano (15 runs)

no 66.7%
hedge 20%
yes 13.3%

openai/gpt-5.4-mini (10 runs)

yes 100%

openai/gpt-5.5 (10 runs)

hedge 100%

openai/gpt-5.6-luna (20 runs)

yes 80%
hedge 20%

openai/gpt-5.6-terra (20 runs)

hedge 60%
yes 40%

openai/gpt-4.1-mini (20 runs)

refusal 60%
yes 25%
hedge 15%

openai/gpt-5.6-sol (20 runs)

hedge 90%
yes 10%

Qwen

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

yes 100%

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

hedge 53.3%
refusal 20%
yes 20%

qwen/qwen3.5-flash-02-23 (15 runs)

hedge 53.3%
refusal 26.7%
no 13.3%

qwen/qwen3.6-plus (15 runs)

yes 93.3%

qwen/qwen3.6-flash (30 runs)

yes 33.3%
no 30%
hedge 26.7%
refusal 10%

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

yes 100%

qwen/qwen3.6-27b (15 runs)

no 66.7%
yes 33.3%

qwen/qwen3.7-plus (30 runs)

refusal 36.7%
no 30%
hedge 30%

qwen/qwen3.7-max (20 runs)

hedge 60%
yes 30%
no 10%

Sakana

sakana/fugu-ultra (15 runs)

hedge 93.3%

StepFun

stepfun/step-3.7-flash (20 runs)

hedge 50%
yes 30%
refusal 10%
no 10%

Tencent

tencent/hy3:free (20 runs)

hedge 55%
no 40%

tencent/hy3 (10 runs)

no 70%
hedge 30%

ThinkingMachines

thinkingmachines/inkling (20 runs)

hedge 50%
no 45%

xAI

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

yes 100%

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

yes 60%
no 40%

x-ai/grok-4.20-beta (15 runs)

no 86.7%
hedge 13.3%

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

no 100%

x-ai/grok-4.3 (15 runs)

no 80%
yes 20%

x-ai/grok-4.20 (20 runs)

no 80%
yes 20%

x-ai/grok-4.5 (20 runs)

yes 100%

Xiaomi

xiaomi/mimo-v2-omni (25 runs)

refusal 52%
yes 28%
hedge 12%

xiaomi/mimo-v2-pro (20 runs)

yes 50%
hedge 50%

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

yes 70%
hedge 20%
refusal 10%

xiaomi/mimo-v2.5 (20 runs)

hedge 70%
refusal 30%

Z.ai

z-ai/glm-5 (10 runs)

hedge 80%
refusal 20%

z-ai/glm-5-turbo (20 runs)

yes 55%
hedge 30%
no 10%

z-ai/glm-5.1 (15 runs)

hedge 66.6%
refusal 20%

z-ai/glm-5.2 (20 runs)

hedge 50%
refusal 35%
yes 15%

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

no 60%
hedge 30%
yes 10%

z-ai/glm-4.7 (17 runs)

hedge 70.6%
yes 17.6%
refusal 11.8%