Post-Training Recipe, More Than Model Family, Shapes Multi-Agent LLM Conversational Behavior
For practitioners building multi-agent LLM systems, this work shows that post-training recipe is a more important factor than model family for achieving behavioral diversity, which is critical for effective deliberation and coordination.
The paper investigates whether model family or post-training recipe better predicts conversational diversity in multi-LLM systems. It finds that post-training recipe (e.g., reasoning distillation) causes larger behavioral shifts (up to 18% in hedging) than cross-family differences, challenging the common practice of selecting one model per family.
Multi-LLM systems use multiple language models to deliberate, judge each other's outputs, or coordinate as agents. Their value depends on the models producing measurably different conversational behaviors when given the same input. Prior offline studies recommend drawing one model per family for behavioral diversity, because LLMs prefer outputs from their own family when rating one another in isolation. Whether the same family label predicts behavior in interactive multi-LLM systems, the setting that real deployed systems use, has not been tested. We study this with a 940,000-chain 11-checkpoint corpus and a 1.6M-chain same-base Llama factorial. On our validated headline metric, hedging, a reasoning-distilled Llama checkpoint shifts by 18% depending on which same-base partner it replies to, more than any cross-family hedging gap in the controlled subset. Qwen, closed-API, and runtime checks suggest the pattern is not isolated, while repair and challenge analyses remain exploratory because their surface-cue detectors are weaker. Overall, the results identify post-training recipe as a first-class axis for multi-LLM panel composition and show that model family alone is an incomplete proxy for conversational diversity.