We evaluated the efficacy and safety of short‐term fully closed‐loop insulin delivery using faster versus standard insulin aspart in type 2 diabetes. Fifteen adults with insulin‐treated type 2 diabetes underwent 22 hours of closed‐loop insulin delivery with either faster or standard insulin aspart in a double‐blind randomised crossover design. Basal‐bolus regimen was replaced by model predictive control algorithm‐directed insulin delivery based on sensor glucose levels. The primary outcome was time with plasma glucose in target range (5.6‐10.0mmol/l) and did not differ between treatments (mean difference [95%CI] ‐3.3% [8.2;1.7], p=0.17). Mean glucose and glucose variability were comparable, as was time spent below and above target range. Hypoglycaemia (<3.5mmol/l) occurred once with faster insulin aspart and twice with standard insulin aspart. Mean total insulin dose was higher with faster insulin aspart (mean difference [95%CI] 3.7U [0.7;6.8], p=0.021). No episodes of severe hypoglycaemia or other serious adverse events occurred. In conclusion, short‐term fully closed‐loop in type 2 diabetes may require higher dose of faster insulin aspart compared to standard insulin aspart to achieve comparable glucose control.