According to God of Prompt on Twitter, recent tests show that as reasoning token count increases during simple counting tasks with distractors, Claude's accuracy drops due to heightened sensitivity to irrelevant information. In contrast, OpenAI's o-series models maintain focus but tend to overfit to the specific problem framing, rather than being distracted. This highlights a divergence in reasoning approaches between leading AI models, with implications for task reliability and practical deployment in business applications that require consistent accuracy in data processing and reasoning under noise (source: God of Prompt, Twitter, Jan 8, 2026).