BenchTable - Reasoning: leaderboard
Dubesor's small-scale personal benchmark, category score for the 30 reasoning, logic and critical-thinking tasks: difficulty-weighted points (pass, refine, fail, refusal) divided by the best points achieved on each task across models. Percent scale, higher is better; scores can be negative. The benchmark is archived, so the table no longer changes.
Metric: Weighted Score (%). Source: dubesor.de. Saturation forecast: Around January 2027. 347 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 93.3 | #71 |
| 2 | Gemini 3.1 Pro (Preview) | 91.7 | #64 |
| 3 | Claude Opus 4 | 91.5 | #170 |
| 4 | Claude Opus 4 (Thinking) | 91.4 | #170 (Claude Opus 4) |
| 5 | Claude Opus 4.1 (Thinking) | 90.4 | #146 (Claude Opus 4.1) |
| 6 | Claude Sonnet 4.6 | 89.7 | #94 |
| 7 | Claude Opus 4.6 | 88.4 | #67 |
| 8 | Gemini 2.5 Pro | 87.8 | #159 |
| 9 | GLM-5 (Thinking) | 87.6 | #149 (GLM-5) |
| 10 | Claude Sonnet 4 | 87.5 | #207 |
| 11 | Claude Opus 4.1 | 86.2 | #146 |
| 12 | Claude Sonnet 4.5 | 85.7 | #148 |
| 13 | Gemini 2.5 Pro (Preview 03-25) | 85.7 | #126 |
| 14 | Claude Sonnet 4 (Thinking) | 85.6 | #207 (Claude Sonnet 4) |
| 15 | Qwen 3 Max | 85.5 | #213 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=benchtable-reasoning · How It Works · Data refreshed daily, snapshot 2026-11-05.