feat: add glm 5 to core (#9436)

* feat: add glm 5 to core

* test glm 5

letta/constants.py

@@ -252,8 +252,11 @@ LLM_MAX_CONTEXT_WINDOW = {
     "deepseek-chat": 64000,
     "deepseek-reasoner": 64000,
     # glm (Z.AI)
-    "glm-4.6": 200000,
     "glm-4.5": 128000,
+    "glm-4.6": 200000,
+    "glm-4.7": 200000,
+    "glm-5": 200000,
+    "glm-5-code": 200000,
     ## OpenAI models: https://platform.openai.com/docs/models/overview
     # gpt-5
     "gpt-5": 272000,

letta/llm_api/zai_client.py

@@ -17,7 +17,12 @@ from letta.settings import model_settings
 
 def is_zai_reasoning_model(model_name: str) -> bool:
     """Check if the model is a ZAI reasoning model (GLM-4.5+)."""
-    return model_name.startswith("glm-4.5") or model_name.startswith("glm-4.6") or model_name.startswith("glm-4.7")
+    return (
+        model_name.startswith("glm-4.5")
+        or model_name.startswith("glm-4.6")
+        or model_name.startswith("glm-4.7")
+        or model_name.startswith("glm-5")
+    )
 
 
 class ZAIClient(OpenAIClient):

letta/model_specs/model_prices_and_context_window.json

@@ -29290,6 +29290,36 @@
     "supports_vision": true,
     "supports_web_search": true
   },
+  "zai/glm-5": {
+    "cache_creation_input_token_cost": 0,
+    "cache_read_input_token_cost": 2e-7,
+    "input_cost_per_token": 1e-6,
+    "output_cost_per_token": 3.2e-6,
+    "litellm_provider": "zai",
+    "max_input_tokens": 200000,
+    "max_output_tokens": 128000,
+    "mode": "chat",
+    "supports_function_calling": true,
+    "supports_prompt_caching": true,
+    "supports_reasoning": true,
+    "supports_tool_choice": true,
+    "source": "https://docs.z.ai/guides/overview/pricing"
+  },
+  "zai/glm-5-code": {
+    "cache_creation_input_token_cost": 0,
+    "cache_read_input_token_cost": 3e-7,
+    "input_cost_per_token": 1.2e-6,
+    "output_cost_per_token": 5e-6,
+    "litellm_provider": "zai",
+    "max_input_tokens": 200000,
+    "max_output_tokens": 128000,
+    "mode": "chat",
+    "supports_function_calling": true,
+    "supports_prompt_caching": true,
+    "supports_reasoning": true,
+    "supports_tool_choice": true,
+    "source": "https://docs.z.ai/guides/overview/pricing"
+  },
   "zai/glm-4.7": {
     "cache_creation_input_token_cost": 0,
     "cache_read_input_token_cost": 1.1e-7,

letta/schemas/llm_config.py

@@ -473,7 +473,10 @@ class LLMConfig(BaseModel):
     @classmethod
     def is_zai_reasoning_model(cls, config: "LLMConfig") -> bool:
         return config.model_endpoint_type == "zai" and (
-            config.model.startswith("glm-4.5") or config.model.startswith("glm-4.6") or config.model.startswith("glm-4.7")
+            config.model.startswith("glm-4.5")
+            or config.model.startswith("glm-4.6")
+            or config.model.startswith("glm-4.7")
+            or config.model.startswith("glm-5")
         )
 
     @classmethod
@@ -499,7 +502,7 @@ class LLMConfig(BaseModel):
         if "gemini" in model:
             return True
         # ZAI GLM reasoning models
-        if "glm-4.5" in model or "glm-4.6" in model or "glm-4.7" in model:
+        if "glm-4.5" in model or "glm-4.6" in model or "glm-4.7" in model or "glm-5" in model:
             return True
         # DeepSeek reasoning models
         if "deepseek-r1" in model or "deepseek-reasoner" in model:

letta/schemas/providers/zai.py

@@ -16,6 +16,8 @@ MODEL_CONTEXT_WINDOWS = {
     "glm-4.5": 128000,
     "glm-4.6": 200000,
     "glm-4.7": 200000,
+    "glm-5": 200000,
+    "glm-5-code": 200000,
 }
 
 

tests/integration_test_send_message_v2.py

@@ -37,7 +37,7 @@ all_configs = [
     "openai-gpt-5.json",
     "claude-4-5-sonnet.json",
     "gemini-2.5-pro.json",
-    "zai-glm-4.6.json",
+    "zai-glm-5.json",
 ]
 
 

tests/model_settings/zai-glm-5.json

@@ -0,0 +1,13 @@
+{
+  "handle": "zai/glm-5",
+  "model_settings": {
+    "provider_type": "zai",
+    "temperature": 1.0,
+    "max_output_tokens": 4096,
+    "parallel_tool_calls": false,
+    "thinking": {
+      "type": "enabled",
+      "clear_thinking": false
+    }
+  }
+}