diff --git a/src/axolotl/utils/schemas/peft.py b/src/axolotl/utils/schemas/peft.py
index 1d1d3e746..4cf8dafe7 100644
--- a/src/axolotl/utils/schemas/peft.py
+++ b/src/axolotl/utils/schemas/peft.py
@@ -96,7 +96,7 @@ class LoraConfig(BaseModel):
 
             else:
                 if self.quantization:
-                    if self.quantization.bits >= 4:
+                    if self.quantization.bits > 4:
                         raise ValueError("Can't load qlora in >4 bit")
 
                     if self.quantization.backend == "gptq":