fix: validate_environment returns false all-clear for zai when ZAI_API_KEY unset (fixes #32962)

Signed-off-by: Aloys Jehwin <aloys.jehwin@sap.com>

litellm/utils.py

@@ -6141,6 +6141,11 @@ def validate_environment(
                 keys_in_environment = True
             else:
                 missing_keys.append("MOONSHOT_API_KEY")
+        elif custom_llm_provider == "zai":
+            if "ZAI_API_KEY" in os.environ:
+                keys_in_environment = True
+            else:
+                missing_keys.append("ZAI_API_KEY")
     else:
         ## openai - chatcompletion + text completion
         if (

tests/local_testing/test_validate_environment_zai.py

@@ -0,0 +1,35 @@
+# SPDX-License-Identifier: MIT
+# Tests for validate_environment fix: zai provider was missing from the
+# elif chain in litellm/utils.py, causing a false all-clear when ZAI_API_KEY
+# is unset (fixes #32962).
+
+import os
+import sys
+
+import pytest
+
+sys.path.insert(0, os.path.abspath(os.path.join(os.path.dirname(__file__), "..", "..")))
+
+import litellm
+
+
+def test_zai_missing_key_reports_missing(monkeypatch):
+    monkeypatch.delenv("ZAI_API_KEY", raising=False)
+    result = litellm.validate_environment(model="zai/glm-5.1")
+    assert result["keys_in_environment"] is False
+    assert "ZAI_API_KEY" in result["missing_keys"]
+
+
+def test_zai_key_present_reports_ok(monkeypatch):
+    monkeypatch.setenv("ZAI_API_KEY", "test-key-123")
+    result = litellm.validate_environment(model="zai/glm-5.1")
+    assert result["keys_in_environment"] is True
+    assert "ZAI_API_KEY" not in result["missing_keys"]
+
+
+def test_moonshot_missing_key_still_reports_missing(monkeypatch):
+    """Regression: moonshot neighbour branch still works after the zai addition."""
+    monkeypatch.delenv("MOONSHOT_API_KEY", raising=False)
+    result = litellm.validate_environment(model="moonshot/kimi-k2")
+    assert result["keys_in_environment"] is False
+    assert "MOONSHOT_API_KEY" in result["missing_keys"]