PerfReasoning - High Effort: leaderboard
PerfReasoning both-exact pass rate at high reasoning effort. Higher is better. Source: arxiv.org/abs/2609.04476.
Metric: Both-Exact Pass Rate (%). Source: arxiv.org. Saturation forecast: Around February 2027. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 | 88.4 | #33 |
| 2 | GPT-5.6 Sol | 88.4 | #22 |
| 3 | Claude Opus 4.8 | 43.5 | #39 |
| 4 | Claude Opus 4.7 | 35.6 | #51 |
| 5 | Gemini 3.5 Flash | 35.6 | #66 |
| 6 | Gemini 3.1 Pro (Preview) | 22.5 | #64 |
| 7 | GPT-OSS-20B | 13 | #523 |
| 8 | GPT-5.6 Terra | 3.7 | #40 |
| 9 | DeepSeek V4 Pro | 3.2 | #112 |
| 10 | GPT-OSS-120B | 0 | #346 |
| 11 | Claude Opus 4.6 | 0 | #67 |
| 12 | GPT-5.6 Luna | 0 | #79 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=perfreasoning-high-effort · How It Works · Data refreshed daily, snapshot 2026-11-05.