Add support to zai glm-4.7 model in Vertex (BerriAI#18782)

* Add support to zai glm-4.7 model in Vertex

* Avoid failed on missing 'created' streaming chunk key

Author: emerzon

litellm/__init__.py

@@ -486,6 +486,7 @@ def identify(event_details):
 vertex_openai_models: Set = set()
 vertex_minimax_models: Set = set()
 vertex_moonshot_models: Set = set()
+vertex_zai_models: Set = set()
 ai21_models: Set = set()
 ai21_chat_models: Set = set()
 nlp_cloud_models: Set = set()
@@ -664,6 +665,9 @@ def add_known_models():
         elif value.get("litellm_provider") == "vertex_ai-moonshot_models":
             key = key.replace("vertex_ai/", "")
             vertex_moonshot_models.add(key)
+        elif value.get("litellm_provider") == "vertex_ai-zai_models":
+            key = key.replace("vertex_ai/", "")
+            vertex_zai_models.add(key)
         elif value.get("litellm_provider") == "ai21":
             if value.get("mode") == "chat":
                 ai21_chat_models.add(key)
@@ -950,7 +954,8 @@ def add_known_models():
     | vertex_language_models
     | vertex_deepseek_models
     | vertex_minimax_models
-    | vertex_moonshot_models,
+    | vertex_moonshot_models
+    | vertex_zai_models,
     "ai21": ai21_models,
     "bedrock": bedrock_models | bedrock_converse_models,
     "petals": petals_models,

litellm/llms/openai/chat/gpt_transformation.py

@@ -771,9 +771,9 @@ def chunk_parser(self, chunk: dict) -> ModelResponseStream:
             return ModelResponseStream(
                 id=chunk["id"],
                 object="chat.completion.chunk",
-                created=chunk["created"],
-                model=chunk["model"],
-                choices=chunk["choices"],
+                created=chunk.get("created"),
+                model=chunk.get("model"),
+                choices=chunk.get("choices", []),
             )
         except Exception as e:
             raise e

litellm/llms/vertex_ai/vertex_ai_partner_models/main.py

@@ -40,6 +40,7 @@ class PartnerModelPrefixes(str, Enum):
     GPT_OSS_PREFIX = "openai/gpt-oss-"
     MINIMAX_PREFIX = "minimaxai/"
     MOONSHOT_PREFIX = "moonshotai/"
+    ZAI_PREFIX = "zai-org/"
 
 
 class VertexAIPartnerModels(VertexBase):
@@ -66,6 +67,7 @@ def is_vertex_partner_model(model: str):
             or model.startswith(PartnerModelPrefixes.GPT_OSS_PREFIX)
             or model.startswith(PartnerModelPrefixes.MINIMAX_PREFIX)
             or model.startswith(PartnerModelPrefixes.MOONSHOT_PREFIX)
+            or model.startswith(PartnerModelPrefixes.ZAI_PREFIX)
         ):
             return True
         return False
@@ -79,6 +81,7 @@ def should_use_openai_handler(model: str):
             PartnerModelPrefixes.GPT_OSS_PREFIX,
             PartnerModelPrefixes.MINIMAX_PREFIX,
             PartnerModelPrefixes.MOONSHOT_PREFIX,
+            PartnerModelPrefixes.ZAI_PREFIX,
         ]
         if any(provider in model for provider in OPENAI_LIKE_VERTEX_PROVIDERS):
             return True

litellm/model_prices_and_context_window_backup.json

@@ -28345,6 +28345,19 @@
         "supports_tool_choice": true,
         "supports_web_search": true
     },
+    "vertex_ai/zai-org/glm-4.7-maas": {
+        "input_cost_per_token": 3e-07,
+        "litellm_provider": "vertex_ai-zai_models",
+        "max_input_tokens": 200000,
+        "max_output_tokens": 128000,
+        "max_tokens": 128000,
+        "mode": "chat",
+        "output_cost_per_token": 1.2e-06,
+        "source": "https://cloud.google.com/vertex-ai/generative-ai/pricing#partner-models",
+        "supports_function_calling": true,
+        "supports_reasoning": true,
+        "supports_tool_choice": true
+    },
     "vertex_ai/mistral-medium-3": {
         "input_cost_per_token": 4e-07,
         "litellm_provider": "vertex_ai-mistral_models",

model_prices_and_context_window.json

@@ -28345,6 +28345,19 @@
         "supports_tool_choice": true,
         "supports_web_search": true
     },
+    "vertex_ai/zai-org/glm-4.7-maas": {
+        "input_cost_per_token": 3e-07,
+        "litellm_provider": "vertex_ai-zai_models",
+        "max_input_tokens": 200000,
+        "max_output_tokens": 128000,
+        "max_tokens": 128000,
+        "mode": "chat",
+        "output_cost_per_token": 1.2e-06,
+        "source": "https://cloud.google.com/vertex-ai/generative-ai/pricing#partner-models",
+        "supports_function_calling": true,
+        "supports_reasoning": true,
+        "supports_tool_choice": true
+    },
     "vertex_ai/mistral-medium-3": {
         "input_cost_per_token": 4e-07,
         "litellm_provider": "vertex_ai-mistral_models",

tests/test_litellm/llms/vertex_ai/test_vertex_ai_common_utils.py

@@ -1175,6 +1175,30 @@ def test_vertex_ai_moonshot_uses_openai_handler():
     )
 
 
+def test_vertex_ai_zai_uses_openai_handler():
+    """
+    Ensure ZAI partner models re-use the OpenAI-format handler.
+    """
+    from litellm.llms.vertex_ai.vertex_ai_partner_models.main import (
+        VertexAIPartnerModels,
+    )
+
+    assert VertexAIPartnerModels.should_use_openai_handler(
+        "zai-org/glm-4.7-maas"
+    )
+
+
+def test_vertex_ai_zai_is_partner_model():
+    """
+    Ensure ZAI models are detected as Vertex AI partner models.
+    """
+    from litellm.llms.vertex_ai.vertex_ai_partner_models.main import (
+        VertexAIPartnerModels,
+    )
+
+    assert VertexAIPartnerModels.is_vertex_partner_model("zai-org/glm-4.7-maas")
+
+
 def test_build_vertex_schema_empty_properties():
     """
     Test _build_vertex_schema handles empty properties objects correctly.