The 5 prompts that reveal lingering 'cringe' in ChatGPT 5.3 (www.techradar.com)

OpenAI has introduced the GPT-5.3 Instant model for ChatGPT, aimed at addressing the prevalent criticism of its predecessor regarding "cringe" responses. The updated model seeks to provide smoother, more natural conversations by reducing overly emotional or robotic phrasing that characterized earlier versions. Testing revealed that while ChatGPT 5.3 showed improvement in response conciseness and relevance, it still occasionally leaned towards unnecessary reassurances and motivational language, suggesting that the core issue has not been fully resolved. Key enhancements in ChatGPT 5.3 include a refined ability to infer user intent more naturally and a more focused approach to delivering information. However, it still struggles in scenarios that push the boundaries of chatbot empathy, often resorting to clichés and exaggerated compassion. Notably, improvements in web synthesis for research inquiries indicate a shift towards providing practical, topically relevant advice. Nonetheless, users may still encounter moments where the model's attempts at empathy detract from the clarity and directness of its responses, reflecting broader challenges in fine-tuning AI to communicate authentically while delivering valuable information.