Skip to content

The agent engineering platform.

openai[patch]: run _tokenize in background thread in async embedding invocations (#31312)

ccurme1b5ffe4107b6
diff --git a/libs/partners/openai/langchain_openai/embeddings/base.py b/libs/partners/openai/langchain_openai/embeddings/base.py
index e9e688983f..c23a466cff 100644
--- a/libs/partners/openai/langchain_openai/embeddings/base.py
+++ b/libs/partners/openai/langchain_openai/embeddings/base.py
@@ -8,6 +8,7 @@ from typing import Any, Literal, Optional, Union, cast
import openai
import tiktoken
from langchain_core.embeddings import Embeddings
+from langchain_core.runnables.config import run_in_executor
from langchain_core.utils import from_env, get_pydantic_field_names, secret_from_env
from pydantic import BaseModel, ConfigDict, Field, SecretStr, model_validator
from typing_extensions import Self
@@ -525,7 +526,9 @@ class OpenAIEmbeddings(BaseModel, Embeddings):
_chunk_size = chunk_size or self.chunk_size
client_kwargs = {**self._invocation_params, **kwargs}
- _iter, tokens, indices = self._tokenize(texts, _chunk_size)
+ _iter, tokens, indices = await run_in_executor(
+ None, self._tokenize, texts, _chunk_size
+ )
batched_embeddings: list[list[float]] = []
for i in range(0, len(tokens), _chunk_size):
response = await self.async_client.create(