diff --git a/modelopt/torch/quantization/model_calib.py b/modelopt/torch/quantization/model_calib.py
index 661f429b2f5..7a01c3f4559 100644
--- a/modelopt/torch/quantization/model_calib.py
+++ b/modelopt/torch/quantization/model_calib.py
@@ -1161,10 +1161,21 @@ def awq(
         model: Model to be calibrated.
         forward_loop: A callable which takes the model as argument and
             forwards calibration data through the model.
+        algorithm: The AWQ variant to run; one of ``"awq_lite"``,
+            ``"awq_clip"``, or ``"awq_full"``.
+
+    Raises:
+        ValueError: If ``algorithm`` is not one of the valid AWQ algorithms.
 
     See :class:`AWQFullCalibConfig <modelopt.torch.quantization.config.AWQFullCalibConfig>` for
     details on the remaining arguments.
     """
+    if algorithm not in ["awq_lite", "awq_clip", "awq_full"]:
+        raise ValueError(
+            f"Invalid AWQ algorithm {algorithm!r}. "
+            "Valid algorithms are 'awq_lite', 'awq_clip', and 'awq_full'."
+        )
+
     with SequentialQuantizer.convert_to_single_quantizer(model):
         if algorithm in ["awq_full", "awq_lite"]:
             awq_lite(model, forward_loop, **kwargs)
diff --git a/tests/unit/torch/quantization/test_calib.py b/tests/unit/torch/quantization/test_calib.py
index 9e58185e383..1aaaabe7605 100644
--- a/tests/unit/torch/quantization/test_calib.py
+++ b/tests/unit/torch/quantization/test_calib.py
@@ -26,6 +26,7 @@
 from modelopt.torch.quantization.config import QuantizerAttributeConfig
 from modelopt.torch.quantization.model_calib import (
     apply_pre_quant_scale_and_smooth,
+    awq,
     disable_pre_quant_scale_and_resmooth,
     layerwise_calibrate,
 )
@@ -391,6 +392,19 @@ def _forward_loop(m):
     )
 
 
+def test_awq_unknown_algorithm_raises():
+    """awq() must fail fast on an unknown algorithm instead of silently skipping calibration."""
+    model = _SimpleMLP()
+
+    with pytest.raises(ValueError, match="Invalid AWQ algorithm 'awq-lite'"):
+        awq(model, algorithm="awq-lite")
+
+    # Valid algorithm strings still go through: awq_lite without a forward_loop
+    # warns and skips instead of raising.
+    with pytest.warns(UserWarning, match="forward_loop must be provided"):
+        awq(model, forward_loop=None, algorithm="awq_lite")
+
+
 def test_smoothquant_enable_disable():
     torch.manual_seed(1234)
     model = _SimpleMLP()