LangChain

LangChain's ChatOpenAI accepts a custom base URL. Point it at PromptsForLess and keep it on Chat Completions.

Install and configureCopy Anchor Link

Install langchain-openai in your Python environment:

pip install langchain-openai

Store your key from the dashboard as PFL_API_KEY, then create the client:

import os
from langchain_openai import ChatOpenAI

llm = ChatOpenAI(
    model="deepseek-v4.1-flash",
    base_url="https://api.promptsforless.com/v1",
    api_key=os.environ["PFL_API_KEY"],
    use_responses_api=False,
)

reply = llm.invoke("Write one sentence about the ocean.")
print(reply.content)
print(reply.usage_metadata)

Swap in any model ID from the catalog. LangChain's ChatOpenAI guide documents the base URL setting and notes that it drops non-standard response fields.

Stream a responseCopy Anchor Link

for chunk in llm.stream("Write a short greeting."):
    print(chunk.content, end="", flush=True)
print()

Add tools and structured outputCopy Anchor Link

PromptsForLess supports tool calls and structured outputs, so LangChain's bind_tools and with_structured_output work as usual. Pick a model with tool calling for agents.

Keep use_responses_api=False. Responses-only LangChain features, such as built-in web search, file search and conversation IDs, are outside this setup.

TroubleshootingCopy Anchor Link

Find model IDs and prices in the catalog.