Information Retrieval on MiddleMatch
13.3Accuracy Improvement (Δ)Ultra-Dense Prompting
Evaluation results
| Method | Links | |
|---|---|---|
| Ultra-Dense PromptingModel=Gemini 2.02026.04 | 13.3 | |
| Ultra-Dense PromptingModel=Claude 3.72026.04 | 11.5 | |
| Ultra-Dense PromptingModel=Average across Models2026.04 | 11.4 | |
| Ultra-Dense PromptingModel=GPT-4o2026.04 | 10.9 | |
| Ultra-Dense PromptingModel=GPT-4o-mini2026.04 | 10.1 |