langchain-predictionguard

Migrating

The Prediction Guard API is compatible with both OpenAI-style and Anthropic-style clients, so the standard LangChain integrations work with it directly. Use whichever matches the functionality you need, pointed at the Prediction Guard API with your existing API key.

Chat (OpenAI-compatible)

pip install langchain-openai

import os

from langchain_openai import ChatOpenAI

chat = ChatOpenAI(
    model="<model-name>",
    api_key=os.environ["PREDICTIONGUARD_API_KEY"],
    base_url="https://api.predictionguard.com",
)

chat.invoke("Tell me a joke")

Chat (Anthropic-compatible)

pip install langchain-anthropic

import os

from langchain_anthropic import ChatAnthropic

chat = ChatAnthropic(
    model="<model-name>",
    api_key=os.environ["PREDICTIONGUARD_API_KEY"],
    base_url="https://api.predictionguard.com",
)

chat.invoke("Tell me a joke")

Completions

import os

from langchain_openai import OpenAI

llm = OpenAI(
    model="<model-name>",
    api_key=os.environ["PREDICTIONGUARD_API_KEY"],
    base_url="https://api.predictionguard.com",
)

llm.invoke("Tell me a joke about bears")

Embeddings

import os

from langchain_openai import OpenAIEmbeddings

embeddings = OpenAIEmbeddings(
    model="<model-name>",
    api_key=os.environ["PREDICTIONGUARD_API_KEY"],
    base_url="https://api.predictionguard.com",
    # Send raw text rather than OpenAI tiktoken token IDs.
    check_embedding_ctx_length=False,
)

embeddings.embed_query("This is an embedding example.")

For the full list of endpoints and models, see the Prediction Guard documentation.