ReLE - Coding: leaderboard
Metric: Accuracy (%). Source: raw.githubusercontent.com. 98 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3.6 Max Preview | 75.7 |
| 2 | GPT-5.6 Pro Sol | 75 |
| 3 | GPT-6 | 74.8 |
| 4 | Gemini 3.7 Flash | 74.8 |
| 5 | Gemini 3.1 Pro (Preview) | 74.2 |
| 6 | Gemini 3.5 Flash | 74.2 |
| 7 | Grok 4.6 | 74.2 |
| 8 | Qwen 3.7 Max | 73.3 |
| 9 | Gemini 3.8 Flash | 72.7 |
| 10 | Claude Opus 5 | 72.5 |
| 11 | DeepSeek V4.1 Flash | 72.5 |
| 12 | Claude Opus 4.8 (Thinking) | 72 |
| 13 | GPT-5.5 | 71.5 |
| 14 | Grok 4.5 | 71.3 |
| 15 | GPT-5.4 (High) | 70.5 |
Interactive version: theaggregate.ai/benchmark?slug=rele-coding · How It Works · Data refreshed daily, snapshot 2026-09-19.