Mirror Zhanlu client: GLM tool_stream, shared v4 UUID request id, drop chat state:ERROR

internal/openai/types.go

@@ -1,6 +1,9 @@
 package openai
 
-import "encoding/json"
+import (
+	"encoding/json"
+	"strings"
+)
 
 type ChatCompletionRequest struct {
 	Model    string                     `json:"model"`
@@ -40,6 +43,11 @@ func (r ChatCompletionRequest) MarshalForUpstream() ([]byte, error) {
 			m[k] = anyValue
 		}
 	}
+	// Mirror the Zhanlu plugin: GLM models require {tool_stream:true} to
+	// stream tool calls back to the client (matches /glm/i.test(model)).
+	if strings.Contains(strings.ToLower(r.Model), "glm") {
+		m["tool_stream"] = true
+	}
 	return json.Marshal(m)
 }