diff --git a/cmd/gguf-parser/main.go b/cmd/gguf-parser/main.go
index b3ddd42..43a2e60 100644
--- a/cmd/gguf-parser/main.go
+++ b/cmd/gguf-parser/main.go
@@ -573,9 +573,12 @@ func mainAction(c *cli.Context) error {
 		eopts = append(eopts, WithinMaxContextSize())
 	}
 	if logicalBatchSize > 0 {
-		eopts = append(eopts, WithLogicalBatchSize(int32(logicalBatchSize)))
+		eopts = append(eopts, WithLogicalBatchSize(int32(max(32, logicalBatchSize))))
 	}
 	if physicalBatchSize > 0 {
+		if physicalBatchSize > logicalBatchSize {
+			return errors.New("--ubatch-size must be less than or equal to --batch-size")
+		}
 		eopts = append(eopts, WithPhysicalBatchSize(int32(physicalBatchSize)))
 	}
 	if parallelSize > 0 {
diff --git a/file_estimate.go b/file_estimate.go
index e446efc..5cf1068 100644
--- a/file_estimate.go
+++ b/file_estimate.go
@@ -103,10 +103,16 @@ func (gf *GGUFFile) EstimateLLaMACppUsage(opts ...LLaMACppUsageEstimateOption) (
 	}
 	if o.LogicalBatchSize == nil {
 		o.LogicalBatchSize = ptr.To(int32(2048))
+	} else {
+		// See https://github.com/ggerganov/llama.cpp/blob/0bf16de07b0692e7df26b9a633e232bbd66e0360/src/llama.cpp#L16519-L16525.
+		o.LogicalBatchSize = ptr.To(max(32, *o.LogicalBatchSize))
 	}
 	if o.PhysicalBatchSize == nil {
 		o.PhysicalBatchSize = ptr.To(int32(512))
 	}
+	if *o.PhysicalBatchSize > *o.LogicalBatchSize {
+		panic("physical batch size must be less than or equal to logical batch size")
+	}
 
 	// Architecture and tokenizer metadata.
 	var (