Has the Creativity of Large-Language Models peaked? An analysis of inter- and intra-LLM variability