LangChain with SovereignEG

Goal: call a SovereignEG model from LangChain, then build a small prompt chain.

Time: 5 minutes.

LangChain's ChatOpenAI class works with any OpenAI-compatible server. You only give it a base_url (https://backend.sovereigneg.com/v1), an api_key, and the model name.

Step 1 — Install

pip install langchain-openai

Step 2 — One chat call

Save as langchain_hello.py:

import os
import time
 
from langchain_openai import ChatOpenAI
 
llm = ChatOpenAI(
    model="SovereignEG/Qwen3.8-27B-FP8",
    base_url="https://backend.sovereigneg.com/v1",
    api_key=os.environ["SOVEREIGNEG_API_KEY"],
)
 
start = time.time()
response = llm.invoke("Hello, who are you?")
elapsed = time.time() - start
 
print(response.content)
print(f"\nTime: {elapsed:.2f}s")
print(f"Tokens: {response.usage_metadata}")

Run it:

python langchain_hello.py

You get the answer, the time it took, and the token count (input_tokens, output_tokens, total_tokens) — this is what you are billed for.

Step 3 — Streaming

Print tokens as they arrive instead of waiting for the full answer:

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

Step 4 — A prompt chain

Chain a prompt template → model → string output with LCEL (the | operator):

from langchain_core.prompts import ChatPromptTemplate
from langchain_core.output_parsers import StrOutputParser
 
prompt = ChatPromptTemplate.from_messages([
    ("system", "You are a helpful assistant. Answer in {language}."),
    ("user", "{question}"),
])
 
chain = prompt | llm | StrOutputParser()
 
print(chain.invoke({"language": "Arabic", "question": "What is the capital of Egypt?"}))

Step 5 — Useful options

llm = ChatOpenAI(
    model="SovereignEG/Qwen3.8-27B-FP8",
    base_url="https://backend.sovereigneg.com/v1",
    api_key=os.environ["SOVEREIGNEG_API_KEY"],
    temperature=0.2,      # lower = more predictable
    max_tokens=512,       # cap the answer length (and cost)
    timeout=60,           # seconds
    max_retries=2,
)

JSON output (the model supports response_format):

json_llm = llm.bind(response_format={"type": "json_object"})
print(json_llm.invoke("Return a JSON object with keys city and country for Cairo.").content)

Troubleshooting

Problem	Fix
`AuthenticationError`	`SOVEREIGNEG_API_KEY` is not set in this terminal
`BadRequestError` … `model_not_found`	Use the exact id from `/v1/models`
Very slow first call	Normal: the first request warms up the connection; later calls are faster

Next: LangGraph