OpenAI chief Sam Altman praised the new GPT-5.4 model as his favorite version to converse with, highlighting improvements in personality and coding ability. He also recognized three shortcomings—frontend aesthetic taste, occasional lapses in real‑world context, and incomplete task execution—that the company plans to address. The remarks underscore OpenAI’s shift toward refining how ChatGPT feels to use, not just its raw performance, as it competes with rivals such as Claude, Gemini and Opus. Read more →