As global temperatures rise, the mechanistic links between temperature, physiology and behaviour will increasingly define predictions of ecological change. However, for many taxa, we currently lack consensus about how thermal performance traits vary within and across populations, and whether and how locally adaptive trait plasticity can buffer warming effects. The metabolic cold adaptation hypothesis posits that cold environments (e.g. high elevations and latitudes) select for high metabolic rates (MR), even after controlling for body size differences, and that this enables high activity levels when an organism is near its cold lower thermal limits. Steep MR reaction norms are further predicted at cold temperatures to enable rapid behavioural activation with rising temperatures needed to exploit brief thermal windows suitable for performing eco-evolutionary tasks. We tested these predictions by performing common garden experiments comparing thermal reaction norms of MR (from 15 to 32°C) and behaviour (from 10 to 40°C) across populations of the ant Aphaenogaster iberica sampled from a 2 km elevation gradient in the Sierra Nevada Mountains of southern Spain. As predicted, high-elevation ants had higher MR and steeper MR-temperature reaction norms. However, higher rates of energy use did not yield the predicted benefits of steeper activity-level reaction norms. The evidence for locally adaptive metabolic physiology only became apparent at intermediate temperatures, highlighting the importance of testing thermal performance hypotheses across thermal gradients, rather than focusing only on performance at thermal limits (i.e. critical thermal values). The partial support for the metabolic cold adaptation hypothesis highlights that while organisms likely show a wealth of unexplored metabolic temperature plasticity, the physiological mechanisms and eco-evolutionary trade-offs underlying such local adaptation remain obscure.