NYT Connections Extended: leaderboard
Extended version: 940 puzzles with extra trick words added that don't fit any category. Models get one attempt. Significantly harder than the original.
Metric: Score (%). Source: github.com. Status: saturation imminent. 104 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 96.1 |
| 2 | GPT-5.5 (xHigh) | 96.1 |
| 3 | GPT-5.5 (High) | 95.2 |
| 4 | GPT-5.4 (xHigh) | 93.4 |
| 5 | Gemini 3 Pro (Preview) | 92.3 |
| 6 | Kimi K3 | 92.1 |
| 7 | Gemini 3.5 Flash | 90.8 |
| 8 | Qwen 3.8 Max | 88.9 |
| 9 | Gemini 3.6 Flash | 88.5 |
| 10 | GPT-5.4 (High) | 88.4 |
| 11 | Claude Opus 4.6 (Thinking, High) | 88.1 |
| 12 | GPT-5.5 (Medium) | 87.3 |
| 13 | DeepSeek V4 Flash | 87.2 |
| 14 | Kimi K2.6 | 80.2 |
| 15 | Qwen 3.7 Max | 78.9 |
Interactive version: theaggregate.ai/benchmark?slug=nyt-connections-extended · How It Works · Data refreshed daily, snapshot 2026-09-05.