At the time of launch, OpenAI said these models outperformed both GPT-4o and the GPT-4.5 AI models in coding capability. Based on internal testing, the company claimed that GPT-4.1 scored 54.6 percent ...