Add LiteLLM AI gateway provider for LLM and embeddings

pyproject.toml

@@ -52,6 +52,7 @@ anthropic = ["anthropic>=0.49.0,<0.50.0"]
 bedrock = ["boto3>=1.35.0,<2.0.0"]
 ollama = ["ollama>=0.4.4,<0.5.0"]
 openai = ["openai>=1.51.1,<2.0.0"]
+litellm = ["litellm>=1.80.0,<1.87.0"]
 mistralai = ["mistralai>=1.0.3,<2.0.0"]
 qdrant = ["qdrant-client>=1.11.3,<2.0.0"]
 kg_creation_tools = ["neo4j-viz>=0.4.2,<0.5.0"]

src/neo4j_graphrag/embeddings/__init__.py

@@ -16,6 +16,7 @@
 from .bedrock import BedrockEmbeddings
 from .cohere import CohereEmbeddings
 from .google_genai import GeminiEmbedder
+from .litellm import LiteLLMEmbeddings
 from .mistral import MistralAIEmbeddings
 from .ollama import OllamaEmbeddings
 from .openai import AzureOpenAIEmbeddings, OpenAIEmbeddings
@@ -33,4 +34,5 @@
     "MistralAIEmbeddings",
     "CohereEmbeddings",
     "GeminiEmbedder",
+    "LiteLLMEmbeddings",
 ]

src/neo4j_graphrag/embeddings/litellm.py

@@ -0,0 +1,116 @@
+#  Copyright (c) "Neo4j"
+#  Neo4j Sweden AB [https://neo4j.com]
+#  #
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#  #
+#      https://www.apache.org/licenses/LICENSE-2.0
+#  #
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+
+from __future__ import annotations
+
+from typing import Any, Optional
+
+from neo4j_graphrag.embeddings.base import Embedder
+from neo4j_graphrag.exceptions import EmbeddingsGenerationError
+from neo4j_graphrag.utils.rate_limit import (
+    RateLimitHandler,
+    async_rate_limit_handler,
+    rate_limit_handler,
+)
+
+
+class LiteLLMEmbeddings(Embedder):
+    """Embeddings via LiteLLM AI gateway.
+
+    LiteLLM provides a unified interface to 100+ embedding providers using the
+    OpenAI embedding format.
+
+    Args:
+        model (str): The embedding model identifier in LiteLLM format
+            (e.g. "text-embedding-ada-002", "cohere/embed-english-v3.0").
+        rate_limit_handler (Optional[RateLimitHandler]): A rate limit handler
+            to manage API rate limits. Defaults to None.
+        **kwargs (Any): Arguments passed to ``litellm.embedding``
+            (e.g. ``api_key``, ``api_base``).
+
+    Raises:
+        EmbeddingsGenerationError: If there's an error generating embeddings.
+
+    Example:
+
+    .. code-block:: python
+
+        from neo4j_graphrag.embeddings import LiteLLMEmbeddings
+
+        embedder = LiteLLMEmbeddings(model="text-embedding-ada-002", api_key="...")
+        vector = embedder.embed_query("Some text")
+    """
+
+    def __init__(
+        self,
+        model: str = "text-embedding-ada-002",
+        rate_limit_handler: Optional[RateLimitHandler] = None,
+        **kwargs: Any,
+    ) -> None:
+        try:
+            import litellm
+        except ImportError:
+            raise ImportError(
+                """Could not import litellm python client.
+                Please install it with `pip install "neo4j-graphrag[litellm]"`."""
+            )
+        super().__init__(rate_limit_handler)
+        self.litellm = litellm
+        self.model = model
+        self.kwargs = kwargs
+
+    @rate_limit_handler
+    def embed_query(self, text: str, **kwargs: Any) -> list[float]:
+        """Generate embeddings for a given query using LiteLLM.
+
+        Args:
+            text (str): The text to generate an embedding for.
+            **kwargs (Any): Additional arguments passed to ``litellm.embedding``.
+        """
+        try:
+            response = self.litellm.embedding(
+                model=self.model,
+                input=[text],
+                **self.kwargs,
+                **kwargs,
+            )
+            embedding: list[float] = response.data[0]["embedding"]
+            return embedding
+        except Exception as e:
+            raise EmbeddingsGenerationError(
+                f"Failed to generate embedding with LiteLLM: {e}"
+            ) from e
+
+    @async_rate_limit_handler
+    async def async_embed_query(self, text: str, **kwargs: Any) -> list[float]:
+        """Asynchronously generate embeddings for a given query using LiteLLM.
+
+        Args:
+            text (str): The text to generate an embedding for.
+            **kwargs (Any): Additional arguments passed to ``litellm.aembedding``.
+        """
+        try:
+            response = await self.litellm.aembedding(
+                model=self.model,
+                input=[text],
+                **self.kwargs,
+                **kwargs,
+            )
+            embedding: list[float] = response.data[0]["embedding"]
+            return embedding
+        except Exception as e:
+            raise EmbeddingsGenerationError(
+                f"Failed to generate embedding with LiteLLM: {e}"
+            ) from e

src/neo4j_graphrag/llm/__init__.py

@@ -20,6 +20,7 @@
 from .bedrock_llm import BedrockLLM
 from .cohere_llm import CohereLLM
 from .google_genai_llm import GeminiLLM
+from .litellm_llm import LiteLLMChat
 from .mistralai_llm import MistralAILLM
 from .ollama_llm import OllamaLLM
 from .openai_llm import AzureOpenAILLM, OpenAILLM
@@ -31,6 +32,7 @@
     "BedrockLLM",
     "CohereLLM",
     "GeminiLLM",
+    "LiteLLMChat",
     "LLMResponse",
     "LLMUsage",
     "LLMBase",