Concepts

Chat Model

Wraps any LangChain-compatible chat model. Supports plain string prompts and full message-list conversations (needed for multi-turn tool calling). Optionally accepts tools=[...] to enable tool calling.

Why it exists

Gives every provider (OpenAI, Groq, Gemini, Anthropic) the same invoke()/stream()/ainvoke() interface, so swapping providers never changes the rest of your chain.

View source on GitHub
from educhain.prompts import PromptTemplate
from educhain.models import ChatModel
from educhain.output_parsers import StringOutputParser

model = ChatModel()  # defaults to ChatOpenAI under the hood

prompt = PromptTemplate(
    template="Explain {topic} in simple words.",
    input_variables=["topic"]
)
parser = StringOutputParser()

chain = prompt | model | parser
response = chain.invoke({"topic": "Artificial Intelligence"})
print(response)

Execution Styles

ChatModel supports three ways to run: invoke() for a single synchronous call, stream() for token-by-token output, and ainvoke() for async execution — all implemented directly on ChatModel itself.

educhain.py
# Synchronous
response = model.invoke("Explain AI in simple words.")

# Streaming
for chunk in model.stream("Explain AI in simple words."):
    print(chunk, end="")

# Async
response = await model.ainvoke("Explain AI in simple words.")

Related