CreativityBench - Entity Correct: leaderboard
CreativityBench (2026) creative tool use, scored leniently: the percent of the 14,000 tasks where the model picks the gold entity, whichever part it names; higher is better.
Metric: Entity correct rate (%, times 100): share of tasks where the model selects the gold entity, whatever the part, over the 14K affordance-based creative tool-use tasks of CreativityBench (eight household scenes; the model sees the task and every entity in the scene with its parts and attributes, and must name the entity and part to repurpose and how to use it), temperature 0 except GPT-5 Mini and Nano (fixed default sampling); higher is better. Source: arxiv.org. Saturation forecast: Not forecast. 10 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3 32B | 62.46 | #972 |
| 2 | Qwen 3 14B | 61.41 | #1269 |
| 3 | GPT-5 Nano | 57.21 | #906 |
| 4 | Llama 3.3 70B Instruct | 55.32 | #1385 |
| 5 | Qwen 3 4B | 52.77 | #2348 |
| 6 | Ministral-3-14B-Reasoning-2512 | 52.63 | #1457 |
| 7 | GPT-5.2 | 52.1 | #185 |
| 8 | GPT-5 Mini | 48.56 | #364 |
| 9 | Gemini 2.5 Flash | 36.94 | #463 |
| 10 | Gemini 2.5 Pro | 35.52 | #276 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=creativitybench-entity-correct · How It Works · Data refreshed daily, snapshot 2026-11-21.