CulMind-R: leaderboard
Metric: Final-answer score (0-100) on the 24-task CulMind-R reasoning subset, the model writing a structured reasoning process before its answer; higher is better. Source: arxiv.org. Saturation forecast: Around 2029. 14 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3 VL 32B Instruct | 64.9 |
| 2 | GPT-5.5 | 63.5 |
| 3 | Gemini 3 Flash (Preview) | 61.3 |
| 4 | Qwen 3.5 27B | 59.6 |
| 5 | GLM-4.5V | 58.3 |
| 6 | GPT-5.4 Mini | 54.6 |
| 7 | Qwen 3 VL 8B Instruct | 54.4 |
| 8 | Qwen 3.5 9B | 53.5 |
| 9 | InternVL3-14B | 51.3 |
| 10 | InternVL3-8B | 50.8 |
| 11 | Gemini 2.5 Flash | 48.2 |
Interactive version: theaggregate.ai/benchmark?slug=culmind-r · How It Works · Data refreshed daily, snapshot 2026-09-29.