diff --git a/src/libstat/classifiers/bayes.c b/src/libstat/classifiers/bayes.c
index dbae98cc2..1d5bb2a6f 100644
--- a/src/libstat/classifiers/bayes.c
+++ b/src/libstat/classifiers/bayes.c
@@ -331,34 +331,58 @@ bayes_classify_token_multiclass(struct rspamd_classifier *ctx,
 
 		w = (fw * total_count) / (1.0 + fw * total_count);
 
-		/* Apply multinomial model for each class */
-		for (j = 0; j < cl->num_classes; j++) {
-			/* Skip classes with insufficient learns */
-			if (ctx->cfg->min_learns > 0 && cl->class_learns[j] < ctx->cfg->min_learns) {
-				continue;
+		if (cl->num_classes == 2) {
+			/* Binary-compatible path: normalize per-token probabilities across the two classes */
+			double f0 = (double) class_counts[0] / MAX(1.0, (double) cl->class_learns[0]);
+			double f1 = (double) class_counts[1] / MAX(1.0, (double) cl->class_learns[1]);
+			double denom = f0 + f1;
+
+			if (denom > 0.0) {
+				double p0 = f0 / denom;
+				double p1 = f1 / denom;
+				double bp0 = PROB_COMBINE(p0, total_count, w, 0.5);
+				double bp1 = PROB_COMBINE(p1, total_count, w, 0.5);
+
+				/* Bound and apply min strength (relative to 0.5 for binary) */
+				bp0 = MAX(0.0, MIN(1.0, bp0));
+				bp1 = MAX(0.0, MIN(1.0, bp1));
+
+				if (fabs(bp0 - 0.5) >= ctx->cfg->min_prob_strength) {
+					cl->class_log_probs[0] += log(bp0);
+				}
+				if (fabs(bp1 - 0.5) >= ctx->cfg->min_prob_strength) {
+					cl->class_log_probs[1] += log(bp1);
+				}
 			}
+		}
+		else {
+			/* General multinomial model for N>2 classes */
+			for (j = 0; j < cl->num_classes; j++) {
+				/* Skip classes with insufficient learns */
+				if (ctx->cfg->min_learns > 0 && cl->class_learns[j] < ctx->cfg->min_learns) {
+					continue;
+				}
 
-			double class_freq = (double) class_counts[j] / MAX(1.0, (double) cl->class_learns[j]);
-			double class_prob = PROB_COMBINE(class_freq, total_count, w, 1.0 / cl->num_classes);
+				double class_freq = (double) class_counts[j] / MAX(1.0, (double) cl->class_learns[j]);
+				double class_prob = PROB_COMBINE(class_freq, total_count, w, 1.0 / cl->num_classes);
 
-			/* Ensure probability is properly bounded [0, 1] */
-			class_prob = MAX(0.0, MIN(1.0, class_prob));
+				/* Ensure probability is properly bounded [0, 1] */
+				class_prob = MAX(0.0, MIN(1.0, class_prob));
 
-			/* Skip probabilities too close to uniform (1/num_classes) */
-			double uniform_prior = 1.0 / cl->num_classes;
-			if (fabs(class_prob - uniform_prior) < ctx->cfg->min_prob_strength) {
-				continue;
-			}
+				/* Skip probabilities too close to uniform (1/num_classes) */
+				double uniform_prior = 1.0 / cl->num_classes;
+				if (fabs(class_prob - uniform_prior) < ctx->cfg->min_prob_strength) {
+					continue;
+				}
 
-			cl->class_log_probs[j] += log(class_prob);
+				cl->class_log_probs[j] += log(class_prob);
+			}
 		}
 
 		cl->processed_tokens++;
 		if (!(tok->flags & RSPAMD_STAT_TOKEN_FLAG_META)) {
 			cl->text_tokens++;
 		}
-
-		/* Per-token debug logging removed to reduce verbosity */
 	}
 }