The study aimed to compare and rank the efficacy of various eating patterns for glycemic control, anthropometrics, and serum lipid profiles in the management of type 2 diabetes and prediabetes, and provide evidence for personalized clinical decision-making. We conducted a network meta-analysis using arm-based Bayesian methods and random effect models following the Cochrane handbook. We drew the conclusions using the partially contextualized framework by the GRADE working group. Twelve English and Chinese databases and registers were retrieved, and we obtained 9,534 references, of which 107 independent studies were eligible, including 8,909 participants, ten experimental diets, and thirteen outcome variables. The meta-analysis denoted that: caloric restriction was ranked as the best pattern for weight loss (SUCRA 86.8%) and reducing waist circumference (82.2%), high-fiber diets for lowering fasting plasma glucose (82.1%) and insulin (79.4%), Dietary Approaches to Stop Hypertension for reducing glycated hemoglobin (90.5%) and systolic blood pressure (87.9%), simple high-protein diets for improving insulin resistance (86.3%) and diastolic blood pressure (74.6%), low-carbohydrate diets for improving body mass index (81.6%) and high-density lipoprotein (84.0%), low-glycemic-index diets for lowering total cholesterol (87.5%) and low-density lipoprotein (86.6%), and Paleolithic diets for reducing triacylglycerol (83.4%). However, the results were of moderate sensitivity, and publication bias of glycated hemoglobin, weight, and body mass index existed. Meta-regression suggested that macronutrients, energy intake, baseline, and weight may modify outcomes differently, while the duration did not show a significant association with results. Forty-nine (39.8%) out of 123 pieces of evidence was rated as moderate quality, and there was no high-quality evidence. Additionally, only 38.2% of the effect sizes of the evidence met the minimally important clinical difference threshold. Clinicians can use the evidence to provide personalized nutrition consultations to patients according to their baseline characteristics. However, the results should be carefully explained and applied because of the sensitivity and low quality.