A user on Reddit shared results from a benchmark called connections-bench, which tests how well AI models can solve the New York Times Connections puzzle. The model 'Fable' performed exceptionally well, with Anthropic's Sonnet 5 and another model, glm-5.2, also showing strong results. OpenAI's GPT-5.5 was nearly on par with Fable, while Haiku struggled, and gpt-4.1-mini performed poorly. AI
IMPACT Provides insight into the puzzle-solving capabilities of various LLMs, highlighting relative strengths and weaknesses.
RANK_REASON User-generated benchmark results shared on Reddit, not an official release or research paper.
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →