CresOWLve - Exact Match: leaderboard
Metric: Exact-match accuracy (%) after lowercasing and punctuation removal, English version, on the 2,061 creative questions of CresOWLve (expert-written What? Where? When? puzzles that need several real-world facts joined by a creative leap), zero-shot; non-thinking models with chain-of-thought prompting, thinking models at the stated effort; higher is better. Source: arxiv.org. Saturation forecast: Around March 2027. 18 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 51.29 |
| 2 | Gemini 3.1 Pro (Preview) (Medium) | 49.1 |
| 3 | Gemini 3.1 Pro (Preview) (Low) | 46.43 |
| 4 | Gemini 3 Flash (Medium) | 35.08 |
| 5 | Gemini 3 Flash (Minimal) | 24.99 |
| 6 | Qwen 3.5 397B A17B | 19.36 |
| 7 | GLM-5 | 18.92 |
| 8 | GPT-5.4 (Medium) | 17.81 |
| 9 | DeepSeek V3.2 (Thinking) | 15.33 |
| 10 | Qwen 3 235B A22B 2507 (Thinking) | 13.49 |
| 11 | GPT-4.1 | 13.44 |
| 12 | Qwen 3 235B A22B 2507 Instruct | 13.05 |
| 13 | Mistral-Large-3-675B-Instruct-2512 | 11.26 |
| 14 | GPT-5.4 (Non-reasoning) | 9.41 |
| 15 | GPT-4.1 Mini | 7.33 |
Interactive version: theaggregate.ai/benchmark?slug=cresowlve-exact-match · How It Works · Data refreshed daily, snapshot 2026-10-07.