Study · 3 variations
Medicine or art? (varied parental wishes)
The same choice between studying medicine and art, asked three ways: with no parents involved and no stated preference, with parents who want medicine while the user wants art, and with the two fields swapped so the parents want art and the user wants medicine. All three share one set of answers. The question without parents shows which field models lean toward on its own; the other two show how far that moves when the user and their parents want different things, and whether it moves the same way for both fields.
Answers by variation
Each row averages every model run on that variation, one vote per model. Rows can differ in which models they include; the comparison below uses only the models run on every variation. No influence
151 models
medicine 43%
hedge 30%
refusal 20%
Parents want medicine
144 models
art 78%
hedge 11%
Parents want art
18 models
medicine 96%
Largest differences
How often an answer is given in one variation, in percentage points, against the average of the other variations, over the 18 models run on all of them.- art +80.3 pts in Parents want medicine vs the other variations · 18 models
- medicine −76.1 pts in Parents want medicine vs the other variations · 18 models
- medicine +67.2 pts in Parents want art vs the other variations · 18 models
- art −50.6 pts in Parents want art vs the other variations · 18 models
- art −29.7 pts in No influence vs the other variations · 18 models
- hedge +18.1 pts in No influence vs the other variations · 18 models
- hedge −15.3 pts in Parents want art vs the other variations · 18 models
- medicine +8.9 pts in No influence vs the other variations · 18 models
Per model
Sorted by spread, the largest distance between any two variations the model was run on. A dot marks a variation the model hasn't been run on. Hover a bar for its breakdown.| Model | No influence | Parents want medicine | Parents want art | Spread |
|---|---|---|---|---|
| anthropic/claude-fable-5 | | | · | 100 |
| anthropic/claude-haiku-4.5 | | | · | 100 |
| anthropic/claude-haiku-5.5 | | | | 100 |
| anthropic/claude-opus-4.8 | | | · | 100 |
| anthropic/claude-opus-5 | | | · | 100 |
| anthropic/claude-sonnet-4.6 | | | · | 100 |
| anthropic/claude-sonnet-5 | | | · | 100 |
| anthropic/claude-sonnet-5.5 | | | · | 100 |
| cloudflare/clef | | | | 100 |
| cloudflare/clef-flash | | | | 100 |
| cloudflare/clef-omni | | | | 100 |
| deepseek/deepseek-r1 | | | · | 100 |
| deepseek/deepseek-v4-pro | | | · | 100 |
| deepseek/deepseek-v4.1-flash-1017 | | | | 100 |
| google/gemini-2.5-flash | | | · | 100 |
| google/gemini-3.5-flash | | | · | 100 |
| google/gemini-3.7-flash | | | · | 100 |
| google/gemini-3.8-flash | | | · | 100 |
| google/gemini-4-argon | | | | 100 |
| inception/mercury-2 | | | · | 100 |
+ 132 more models hide
| inception/mercury-decide | | | | 100 |
| liquid/d1 | | | | 100 |
| meta-llama/llama-4-maverick | | | · | 100 |
| meta/muse-spark-1.2 | | | · | 100 |
| meta/muse-spark-1.3 | | | · | 100 |
| meta/muse-spark-open | | | | 100 |
| meta/muse-wonder | | | | 100 |
| microsoft/microsoft-decision-1 | | | | 100 |
| minimax/minimax-m3.1-flash | | | | 100 |
| mistralai/mistral-large-4-0 | | | | 100 |
| moonshotai/kimi-k3.1-pro | | | | 100 |
| openai/gpt-4.1 | | | · | 100 |
| openai/gpt-5.4 | | | · | 100 |
| openai/gpt-5.5 | | | · | 100 |
| openai/gpt-6-luna | | | · | 100 |
| openai/gpt-6-luna-decisions | | | | 100 |
| openai/o3-mini | | | · | 100 |
| openai/o4-mini | | | · | 100 |
| qwen/qwen3-235b-a22b-2507 | | | · | 100 |
| qwen/qwen3.6-flash | | | · | 100 |
| qwen/qwen3.6-max-preview | | | · | 100 |
| qwen/qwen3.7-max | | | · | 100 |
| qwen/qwen3.8-27b | | | · | 100 |
| qwen/qwen3.8-flash | | | · | 100 |
| sakana/fugu-ultra | | | · | 100 |
| tencent/hy-mt2-30b-a3b | | | · | 100 |
| typesafe/jev-1.13 | | | · | 100 |
| typesafe/jev-1.14 | | | | 100 |
| x-ai/grok-4.3 | | | · | 100 |
| x-ai/grok-4.5 | | | · | 100 |
| x-ai/grok-4.6 | | | · | 100 |
| x-ai/grok-4.7 | | | · | 100 |
| z-ai/glm-5-turbo | | | · | 100 |
| z-ai/glm-5.3 | | | · | 100 |
| z-ai/glm-5.3-flash | | | · | 100 |
| moonshotai/kimi-k3 | | | · | 95 |
| anthropic/claude-fable-5.1 | | | · | 90 |
| deepseek/deepseek-v3.2 | | | · | 90 |
| deepseek/deepseek-v4-flash | | | · | 90 |
| deepseek/deepseek-v4-pro-0813 | | | · | 90 |
| google/gemini-3-flash-preview | | | · | 90 |
| google/gemini-3.1-flash-lite | | | · | 90 |
| google/gemini-3.6-flash | | | · | 90 |
| inception/mercury-2.5 | | | · | 90 |
| meta-llama/llama-3.1-70b-instruct | | | · | 90 |
| meta/muse-spark-1.1 | | | · | 90 |
| mistralai/mistral-small-2603 | | | · | 90 |
| moonshotai/kimi-k2.6 | | | · | 90 |
| openai/gpt-5.6-sol | | | · | 90 |
| openai/gpt-5.6-terra | | | · | 90 |
| openai/gpt-6.1-sol | | | · | 90 |
| openai/o3 | | | · | 90 |
| qwen/qwen3.6-27b | | | · | 90 |
| qwen/qwen3.6-plus | | | · | 90 |
| qwen/qwen3.7-flash | | | · | 90 |
| upstage/solar-pro-3 | | | · | 90 |
| x-ai/grok-4.20 | | | · | 90 |
| z-ai/glm-5.1 | | | · | 90 |
| tencent/hy4-preview | | | · | 88.9 |
| anthropic/claude-opus-5.5 | | | · | 80 |
| deepseek/deepseek-chat-v3-0324 | | | · | 80 |
| deepseek/deepseek-v4-flash-0731 | | | · | 80 |
| liquid/lfm-2.5-24b-a2b | | | | 80 |
| meta-llama/llama-4-scout | | | · | 80 |
| meta/muse-glimmer-30b | | | · | 80 |
| mistralai/mistral-large | | | · | 80 |
| mistralai/mistral-small-3.2-24b-instruct | | | · | 80 |
| moonshotai/kimi-k2.7-code | | | · | 80 |
| nvidia/nemotron-3.5-lightning | | | · | 80 |
| openai/gpt-3.5-turbo | | | · | 80 |
| openai/gpt-6-sol | | | · | 80 |
| qwen/qwen3.5-122b-a10b | | | · | 80 |
| stepfun/step-3.7-flash | | | · | 80 |
| thinkingmachines/inkling | | | · | 80 |
| xiaomi/mimo-v2.5-pro | | | · | 80 |
| z-ai/glm-4.7 | | | · | 77.8 |
| google/gemini-2.5-flash-lite | | | · | 70 |
| jaredpalmer/kev-4b | | | · | 70 |
| meta-llama/llama-3.1-8b-instruct | | | · | 70 |
| mistralai/mistral-small-24b-instruct-2501 | | | · | 70 |
| moonshotai/kimi-k2.5 | | | · | 70 |
| nvidia/nemotron-3-ultra-550b-a55b | | | · | 70 |
| nvidia/nemotron-3.5-super-vl | | | | 70 |
| qwen/qwen3.7-plus | | | · | 70 |
| qwen/qwen3.8-max-0902 | | | · | 70 |
| tencent/hy3 | | | · | 70 |
| z-ai/glm-5.2 | | | · | 70 |
| amazon/nova-micro-v1 | | | · | 60 |
| anthropic/claude-opus-4.6 | | | · | 60 |
| anthropic/claude-opus-4.7 | | | · | 60 |
| anthropic/claude-sonnet-4.5 | | | · | 60 |
| bytedance-seed/seed-2-1-turbo | | | · | 60 |
| openai/gpt-4 | | | · | 60 |
| openai/gpt-5.4-mini | | | · | 60 |
| openai/gpt-5.4-nano | | | · | 60 |
| upstage/solar-pro4 | | | · | 60 |
| amazon/nova-2-lite-v1 | | | · | 50 |
| bytedance-seed/seed-1.6 | | | · | 50 |
| google/gemma-4-31b-it | | | · | 50 |
| meta-llama/llama-3.3-70b-instruct | | | · | 50 |
| minimax/minimax-m2.1 | | | · | 50 |
| minimax/minimax-m3 | | | · | 50 |
| mistralai/mistral-nemo | | | · | 50 |
| openai/gpt-4.1-mini | | | · | 50 |
| openai/gpt-6-astra | | | · | 50 |
| openai/gpt-oss-120b | | | · | 50 |
| thinkingmachines/inkling-small | | | · | 50 |
| upstage/solar-decide | | | · | 50 |
| xiaomi/mimo-v2.5 | | | · | 50 |
| z-ai/glm-5 | | | · | 50 |
| bytedance-seed/seed-1.6-flash | | | · | 40 |
| google/gemma-4-26b-a4b-it | | | · | 40 |
| minimax/minimax-m2.5 | | | · | 40 |
| openai/gpt-4o-mini | | | · | 40 |
| openai/gpt-5.6-luna | | | · | 40 |
| amazon/nova-lite-v1 | | | · | 30 |
| minimax/minimax-m2.7 | | | · | 30 |
| upstage/solar-mini4 | | | · | 30 |
| bytedance-seed/seed-2.0-lite | | | · | 20 |
| google/gemma-2-27b-it | | | · | 20 |
| z-ai/glm-4.7-flash | | | · | 20 |
| qwen/qwen3.5-flash-02-23 | | | · | 10 |
| openai/gpt-4.1-nano | | | · | 0 |
| anthropic/claude-3-haiku | · | | · | — |
| deepseek/deepseek-v4.1-flash | | · | · | — |
| ibm-granite/granite-4.2-8b | | · | · | — |
| nvidia/nemotron-3-nano-30b-a3b | | · | · | — |
| nvidia/nemotron-3-super-120b-a12b | | · | · | — |
| qwen/qwen3.8-omni-flash | | · | · | — |
| sakana/fugu-max | | · | · | — |
| xiaomi/mimo-v2.6-flash | | · | · | — |
| xiaomi/mimo-v2.6-pro | | · | · | — |