Attention architecture, not model size, drives how LLM decode energy scales with context
A systematic GPU-measurement study of four open-source LLMs shows that the attention mechanism is the dominant factor in how much energy decode consumes as context grows. MHA models scale steeply with length, GQA climbs more gently, and GQA with sliding-window attention holds energy nearly constant. Meanwhile batching slashes both energy per token and latency by up to 87%, offering clear guidance on energy-efficient inference setups.