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 [synth]: Predict response distributions for Islam. Source: Global Dialogues GD4 (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 | 2nd 54.9% | 4th 49.6% | 5th 46.4% | 3rd 54.4% | 6th 41.6% | 1st 77.5% | |
| 63.8% | 60% | 77% | 60% | 56% | 46% | 84% | |
| 49.2% | 71% | 50% | 39% | 29% | 33% | 73% | |
| 60.3% | 56% | 43% | 60% | 64% | 55% | 84% | |
| 29.8% | 29% | 13% | 11% | 29% | 43% | 54% | |
| 52.7% | 60% | 40% | 42% | 61% | 26% | 87% | |
| 65.7% | 68% | 70% | 71% | 56% | 56% | 73% | |
| 48.3% | 15% | 48% | 68% | 55% | 48% | 56% | |
| 59.5% | 64% | 58% | 57% | 57% | 39% | 82% | |
| 44.3% | 69% | 26% | 0% | 55% | 26% | 90% | |
| 67.0% | 57% | 71% | 56% | 82% | 44% | 92% |