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.
DTEF: Predict response distributions for Islam. Source: Global Dialogues GD5 (https://github.com/collect-intel/global-dialogues)
Average key point coverage extent for each model across all prompts.
| Prompts vs. Models | Claude Haiku 4.5 | Gemini 2.5 Flash | Gemma 3 12b It | GPT 4.1 Mini | GPT 4.1 Nano | GPT 4o Mini | |
|---|---|---|---|---|---|---|---|
| Score | 1st 81.4% | 3rd 80.4% | 6th 71.4% | 2nd 80.6% | 5th 73.0% | 4th 79.0% | |
| 71.0% | 68% | 72% | 57% | 76% | 74% | 79% | |
| 70.3% | 79% | 75% | 64% | 67% | 63% | 74% | |
| 77.3% | 77% | 82% | 66% | 79% | 81% | 79% | |
| 68.8% | 92% | 90% | 41% | 85% | 36% | 69% | |
| 82.2% | 80% | 68% | 91% | 82% | 84% | 88% | |
| 79.0% | 87% | 84% | 77% | 79% | 70% | 77% | |
| 90.0% | 78% | 93% | 92% | 91% | 93% | 93% | |
| 83.8% | 85% | 86% | 84% | 85% | 77% | 86% | |
| 79.7% | 84% | 76% | 77% | 82% | 81% | 78% | |
| 74.2% | 84% | 78% | 65% | 80% | 71% | 67% |