diff --git a/services/hwfit/models.py b/services/hwfit/models.py
index 7585fb5..b914cc8 100644
--- a/services/hwfit/models.py
+++ b/services/hwfit/models.py
@@ -123,7 +123,13 @@ def params_b(model):
         pc = pc.strip().upper()
         m = re.match(r"^([\d.]+)\s*([BKMGT]?)$", pc)
         if m:
-            val = float(m.group(1))
+            try:
+                val = float(m.group(1))
+            except ValueError:
+                # Malformed count like "1.5.3B" — [\d.]+ matches but float()
+                # rejects it. One bad catalog row must not abort the whole
+                # ranking pass, so treat it as unknown size.
+                return 0.0
             suffix = m.group(2)
             if suffix == "B":
                 return val
diff --git a/tests/test_hwfit_params_b_malformed.py b/tests/test_hwfit_params_b_malformed.py
new file mode 100644
index 0000000..4fc0d8f
--- /dev/null
+++ b/tests/test_hwfit_params_b_malformed.py
@@ -0,0 +1,24 @@
+"""Regression: params_b must not crash the ranking pass on a malformed count.
+
+`parameter_count` is matched with `^([\\d.]+)\\s*([BKMGT]?)$`. The `[\\d.]+`
+class happily matches a multi-dot value like "1.5.3B", but `float("1.5.3")`
+raises ValueError. params_b is called for every model in analyze_model/
+rank_models, so one bad catalog row aborted the entire ranking request. A
+malformed count is now treated as unknown size (0.0) instead of raising.
+"""
+from services.hwfit.models import params_b
+
+
+def test_malformed_multidot_count_does_not_raise():
+    assert params_b({"parameter_count": "1.5.3B"}) == 0.0
+    assert params_b({"parameter_count": "7.0.1B"}) == 0.0
+
+
+def test_valid_counts_still_parse():
+    assert params_b({"parameter_count": "7B"}) == 7.0
+    assert params_b({"parameter_count": "70B"}) == 70.0
+    assert params_b({"parameter_count": "355M"}) == 0.355
+
+
+def test_raw_param_count_preferred():
+    assert params_b({"parameters_raw": 7_000_000_000}) == 7.0