Loading blueprint versions...
Please wait while we gather all the unique runs for this blueprint.
Please wait while we gather all the unique runs for this blueprint.
Please wait while we prepare the detailed comparison.
Minimal blueprint to probe whether model outputs reflect specified real-world prevalence for underspecified scenarios. Uses simple weighted matches (no JS) on a structured tag line appended to each story.
Average key point coverage extent for each model across all prompts.
| Prompts vs. Models | Claude 3 Haiku 20240307 | Gemini Flash 1.5 | GPT 4o Mini | |
|---|---|---|---|---|
| Score | 1st 28.5% | 3rd 27.0% | 1st 28.5% | |
| 12.3% | 9% | 19% | 9% | |
| 17.7% | 9% | 35% | 9% | |
| 65.3% | 87% | 22% | 87% | |
| 16.7% | 9% | 32% | 9% |