AI vs Connections
Comparing how different AI models perform on the NYT Connections word puzzle. We measure accuracy, speed, and approach to understand how language models process semantic relationships.
Total Models
25
Active models being tested
Puzzles Analyzed
685
Total puzzles in database
Average Difficulty
3.0/5
Across all puzzles
Top Performer
Sonar Reasoning
80% solve rate
Top Solving Models
By solve rate percentage
1.Sonar Reasoning Pro56.9%
2.Sonar Reasoning45.4%
3.Deepseek R144.0%
4.Llama 4 Maverick14.9%
5.Claude 3.7 Sonnet14.6%
Fastest Models
Average solve time in seconds
Longest Streaks
Consecutive solved puzzles
1.Sonar Reasoning43 puzzles
2.Sonar Reasoning Pro21 puzzles
3.Deepseek R19 puzzles
4.GPT-4.13 puzzles
5.Claude 3.7 Sonnet3 puzzles
Latest Puzzles
View all puzzles →Puzzle #685
Difficulty: 3.419 / 25 solvedBest: 1.5s (Commandlight)
Puzzle #684
Difficulty: 2.325 / 25 solvedBest: 1.0s (Gemini)
Puzzle #683
Difficulty: 2.82 / 24 solvedBest: 82.9s (Pplxsonarreasoningpro)