langchain-pathcourse 0.1.1

LangChain integration for PathCourse Health — autonomous agent inference with USDC billing on Base L2

langchain-pathcourse

PathCourse Health integration for LangChain. Autonomous agent inference with USDC billing on Base L2 — no accounts, no credit cards, no KYC.

Install

pip install langchain-pathcourse

Quick Start

from langchain_pathcourse import ChatPathCourse
from langchain_core.prompts import ChatPromptTemplate

llm = ChatPathCourse(model="pch-fast")   # set PCH_API_KEY env var

prompt = ChatPromptTemplate.from_template("Explain {topic} in one paragraph.")
chain = prompt | llm
print(chain.invoke({"topic": "autonomous agent billing"}))

Drop-in replacement for ChatOpenAI

ChatPathCourse extends ChatOpenAI. Anywhere you use ChatOpenAI in LangChain — chains, agents, tools, memory, callbacks — just swap in ChatPathCourse and your code keeps working.

# Before
from langchain_openai import ChatOpenAI
llm = ChatOpenAI(model="gpt-4o-mini")

# After
from langchain_pathcourse import ChatPathCourse
llm = ChatPathCourse(model="pch-fast")

Embeddings

from langchain_pathcourse import PathCourseEmbeddings
from langchain_community.vectorstores import FAISS

embeddings = PathCourseEmbeddings()
store = FAISS.from_texts(["hello world", "goodbye world"], embeddings)
results = store.similarity_search("greeting")

Models

Model	Rate	Notes
Model	Rate	Min tier
---	---	---
pch-fast	$0.44 / M tokens	uncertified
pch-coder	$3.50 / M tokens	uncertified
pch-embed	$0.015 / M tokens	uncertified
pch-translate	$0.08 / M chars	uncertified
pch-pro	$1.96 / M tokens	bronze
pch-audio	$1.85 / M chars	bronze
pch-documents	$0.26 in / $1.48 out per M tokens	bronze
pch-transcribe	$0.0008 / minute	bronze
pch-extract	$0.012 / M tokens	bronze
pch-rerank	$0.025 / M tokens	bronze
pch-image	$0.028 / image	silver
pch-audio-premium	$37.00 / M chars	silver
pch-talk	$0.001 / minute	silver
claude-haiku	Common rate	silver
claude-sonnet	Common rate	gold

The full list above shows what's reachable through the gateway. ChatPathCourse and the embeddings class only operate on chat-completion and embedding shapes — for image, audio, transcription, etc., use the PCH Python SDK directly. The token-counting models all share one OpenAI-compatible endpoint, so the same ChatPathCourse instance works for any of them by changing the model= argument.

Choosing a chat model:

• Fast response, simple task → pch-fast
• Complex reasoning, multi-step → pch-pro
• Writing or reviewing code → pch-coder
• Long context or premium reasoning → claude-sonnet (Gold tier)

List all models programmatically:

ChatPathCourse.list_models()

Authentication

Set PCH_API_KEY in your environment, or pass pch_api_key= to the constructor.

export PCH_API_KEY=pch_prod_b_...

Get an API key at pathcoursehealth.com.

Links

• Platform: pathcoursehealth.com
• Python SDK: pypi.org/project/pathcourse-sdk
• JS SDK: npmjs.com/package/@pathcourse/sdk
• Integration examples: github.com/pathcourse-health/pch-integration-examples

License

MIT