Svmuu News: OpenAI posted on X stating that training GPT-5.6 using GPT-Red has significantly improved its robustness. OpenAI replayed some of the strongest GPT-Red attacks for evaluation; none of these attacks had been encountered by the model during training. GPT-5.6 Sol has become OpenAI’s most robust model to date against prompt injection, with a failure rate six times lower than that of the best production model from four months ago.