An analysis of Large Language Model (LLM) response writing styles has revealed a notable similarity between the outputs of Kimi and Claude.

By examining the textual characteristics of responses, researchers created a 'heat map' to visualize how different LLMs' writing patterns align. This method focuses solely on the words used, disregarding other metadata.

The findings suggest that Kimi and Claude exhibit a higher degree of overlap in their linguistic choices compared to other models analyzed. This could imply shared training data, architectural similarities, or similar fine-tuning approaches that influence their stylistic output.

Further investigation into these shared writing fingerprints could provide deeper insights into the internal workings and development of these advanced AI systems, potentially guiding future model design and comparison.