Part 4/9:
In an effort to position GPT-4.5 as a model designed with creative thinking in mind, OpenAI introduced a new 'Vibes Benchmark.' However, critics are skeptical about the usefulness of this benchmark given its subjective nature. Users report mixed experiences when chatting with the model, with some claiming it exudes 'chill vibes' while others highlight its frequent errors. Despite a reported reduction in hallucination rates, it remains clear that GPT-4.5 still struggles with self-awareness and factual accuracy.