ConflictQA - Text Conflict (Complementary): leaderboard
Metric: Exact match (%) on 430 complementary (answers need both sources) questions when the provided evidence contains conflicting text passages (TextConf); zero-shot with a conflict-aware prompt, the mean of two evidence orders; exact match of the generated answer set; higher is better. Source: arxiv.org. Saturation forecast: Around 2029. 12 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.1 | 55.58 |
| 2 | DeepSeek V3.2 (Non-reasoning) | 51.86 |
| 3 | O3 Mini | 48.6 |
| 4 | Qwen 3 8B | 46.98 |
| 5 | GPT-4o | 44.42 |
| 6 | GPT-3.5 Turbo (0125) | 41.4 |
| 7 | DeepSeek V3.2 (Thinking) | 39.3 |
| 8 | Llama 3.1 70B Instruct | 33.72 |
| 9 | Mistral Large 3 | 28.84 |
| 10 | Llama 3.1 8B Instruct | 21.63 |
Interactive version: theaggregate.ai/benchmark?slug=conflictqa-text-conflict-complementary · How It Works · Data refreshed daily, snapshot 2026-10-07.