GPT-5.2 got worse on Terminal Bench 2.0, so is GPT-5.2 Pro (twitter.com)

Recent assessments reveal that OpenAI's latest versions, GPT-5.2 and GPT-5.2 Pro, have underperformed in the Terminal Bench 2.0 evaluation, a benchmark designed to assess language models' coding and problem-solving capabilities. This disappointing performance raises questions about the models' effectiveness in practical applications, particularly in programming-related tasks where accuracy and efficiency are paramount. The significance of this development lies in the growing expectations for AI language models to not only generate coherent text but also perform complex coding tasks reliably. The results suggest that while advancements have been made in natural language understanding, there may be limitations in handling specialized tasks within programming environments. As the AI/ML community continues to push for improved performance benchmarks, these findings could lead to reevaluations of training methodologies and model architectures, prompting a shift in focus toward enhancing specific skill sets necessary for coding applications.