From b42dedbe56e10d92a70948baa3131ec6ca01f9a8 Mon Sep 17 00:00:00 2001
From: Ishan Gupta <ishangupta.ds@gmail.com>
Date: Tue, 25 Aug 2026 12:32:00 +0530
Subject: [PATCH] feat(providers): add Turing Engine LLM serving runtime
 provider

---
 litellm/llms/turing_engine.py     | 42 +++++++++++++++++++++++++++++++
 tests/test_litellm/test_turing.py | 24 ++++++++++++++++++
 2 files changed, 66 insertions(+)
 create mode 100644 litellm/llms/turing_engine.py
 create mode 100644 tests/test_litellm/test_turing.py

diff --git a/litellm/llms/turing_engine.py b/litellm/llms/turing_engine.py
new file mode 100644
index 00000000000..42d5d679248
--- /dev/null
+++ b/litellm/llms/turing_engine.py
@@ -0,0 +1,42 @@
+"""Turing Engine Chat Completions API Config.
+
+Provides OpenAI-compatible adapter configuration for Turing Engine runtime.
+"""
+
+from typing import Final
+
+from litellm.llms.openai.chat.gpt_transformation import OpenAIGPTConfig
+
+
+class TuringConfig(OpenAIGPTConfig):
+    """Configuration class for Turing Engine OpenAI-compatible serving runtime."""
+
+    max_tokens: int | None = None
+    temperature: int | None = None
+    top_p: int | None = None
+    stream: bool | None = None
+    sparsity_ratio: float | None = None
+    use_svd_kv: bool | None = None
+
+    def __init__(
+        self,
+        max_tokens: int | None = None,
+        temperature: int | None = None,
+        top_p: int | None = None,
+        stream: bool | None = None,
+        sparsity_ratio: float | None = None,
+        use_svd_kv: bool | None = None,
+    ) -> None:
+        locals_: Final = locals().copy()
+        for key, value in locals_.items():
+            if key != "self" and value is not None:
+                setattr(self.__class__, key, value)
+
+        self.__class__._is_base_class = False
+
+    def _get_openai_compatible_provider_info(
+        self, api_base: str | None = None, api_key: str | None = None
+    ) -> tuple[str | None, str | None]:
+        default_base = api_base or "http://localhost:8000/v1"
+        default_key = api_key or "turing-local"
+        return default_base, default_key
diff --git a/tests/test_litellm/test_turing.py b/tests/test_litellm/test_turing.py
new file mode 100644
index 00000000000..6ffdc2652f6
--- /dev/null
+++ b/tests/test_litellm/test_turing.py
@@ -0,0 +1,24 @@
+from litellm.llms.turing_engine import TuringConfig
+
+
+def test_turing_config():
+    config = TuringConfig(
+        temperature=1,
+        sparsity_ratio=0.57,
+        max_tokens=1024,
+        stream=True,
+        use_svd_kv=True,
+    )
+    assert config.temperature == 1
+    assert config.sparsity_ratio == 0.57
+    assert config.max_tokens == 1024
+    assert config.stream is True
+    assert config.use_svd_kv is True
+    base, key = config._get_openai_compatible_provider_info(None, None)
+    assert base == "http://localhost:8000/v1"
+    assert key == "turing-local"
+    base_custom, key_custom = config._get_openai_compatible_provider_info(
+        "http://gpu-node:8000/v1", "secret-key"
+    )
+    assert base_custom == "http://gpu-node:8000/v1"
+    assert key_custom == "secret-key"