[feat] Add SGLANG_TOOL_STRICT_LEVEL for tool-call behavior control (#12423)

Signed-off-by: Xinyuan Tong <xinyuantong.cs@gmail.com>

python/sglang/srt/server_args.py

@@ -27,7 +27,7 @@ from typing import Dict, List, Literal, Optional, Union
 import orjson
 
 from sglang.srt.connector import ConnectorType
-from sglang.srt.environ import envs
+from sglang.srt.environ import ToolStrictLevel, envs
 from sglang.srt.function_call.function_call_parser import FunctionCallParser
 from sglang.srt.lora.lora_registry import LoRARef
 from sglang.srt.parser.reasoning_parser import ReasoningParser
@@ -1686,6 +1686,9 @@ class ServerArgs:
         os.environ["SGLANG_ENABLE_DETERMINISTIC_INFERENCE"] = (
             "1" if self.enable_deterministic_inference else "0"
         )
+        # Set the highest strict level for Kimi K2 tool calls
+        if self.tool_call_parser == "kimi_k2":
+            envs.SGLANG_TOOL_STRICT_LEVEL.set(ToolStrictLevel.PARAMETER)
 
     def _handle_cache_compatibility(self):
         if self.enable_hierarchical_cache and self.disable_radix_cache: