ReasonScape R12: leaderboard
Reasoning benchmark with 12 cognitive tasks at fixed 16k context. Uses Wilson CIs, truncation penalties, and geometric mean for balanced scoring. Evaluates local/quantized models on structured reasoning.
Metric: ReasonScore. Source: reasonscape.com. Saturation forecast: Around June 2027. 101 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemma 4 31B | 927.75 | #200 |
| 2 | GLM-4.7 | 894.43 | #192 |
| 3 | Qwen 3 235B A22B 2507 Instruct | 841.58 | #300 |
| 4 | GLM 4.5 Air | 798.07 | #392 |
| 5 | Hunyuan A13B-Instruct | 691.15 | #666 |
| 6 | Gemma 4 E4B | 609.66 | #767 |
| 7 | GLM-4.7 Flash | 596.81 | #509 |
| 8 | GLM-4.6V | 523.76 | #316 |
| 9 | Gemma 3 27B (IT) | 384.94 | #523 |
| 10 | Gemma 3 12B (IT) | 284.87 | #664 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=reasonscape-r12 · How It Works · Data refreshed daily, snapshot 2026-10-20.