Extended NYT Connections 940-Puzzle Set
Games · 2026-02-02
Full 940-puzzle Extended NYT Connections leaderboard from Lech Mazur's public benchmark repository. Scores are percentage-point performance over all 940 extended puzzles, with partial credit and one attempt per puzzle.
Top models (higher is better)
| Model | Score |
|---|---|
| Gemini 3.1 Pro Preview | 98.4 |
| GPT-5.5 | 97.5 |
| Gemini 3 Pro Preview | 96.3 |
| GPT-5.6 Sol | 96.1 |
| Kimi K3 | 95.8 |
| Gemini 3.5 Flash | 95.2 |
| Claude Fable 5 | 95.1 |
| Opus 4.6 | 94.7 |
| Opus 4.8 | 94.1 |
| GPT-5.4 | 94.0 |
| Grok 4.20 | 93.4 |
| Grok 4.1 Fast | 91.8 |
| Kimi K2.6 | 91.4 |
| Qwen3.7-Max | 89.8 |
| Muse Spark 1.1 | 89.6 |
| GPT-5.2 | 88.6 |
| Gemini 3 Flash Preview | 88.4 |
| Grok 4.5 | 86.5 |
| GPT-5.2 Pro | 85.7 |
| Sonnet 4.6 | 85.7 |
| GPT-5.6 Terra | 84.7 |
| GLM-5.1 | 84.4 |
| Sonnet 5 | 82.8 |
| Qwen3.6-Max-Preview | 82.2 |
| GLM-5 | 81.8 |
| GLM-5.2 | 81.7 |
| Gemma 4 31B IT | 79.5 |
| GPT-5.6 Luna | 78.4 |
| Kimi K2.5 | 78.3 |
| DeepSeek-V4-Pro | 75.7 |
| MiniMax M3 | 75.3 |
| GPT-5.4 Mini | 71.8 |
| Qwen3.6 Plus (2026-04-02) | 71.3 |
| Qwen3.5 397B A17B | 69.3 |
| Grok 4.3 | 67.5 |
| Qwen3.5 122B A10B | 63.6 |
| Opus 4.5 | 62.6 |
| Qwen3.5-27B | 60.8 |
| Step 3.7 Flash | 53.6 |
| DeepSeek-V4-Flash | 53.2 |