tests/test_model_pair_metadata.py

import os
os.environ["MOCK_INFERENCE"] = "true"

from fastapi.testclient import TestClient
from app import app

client = TestClient(app)

def test_model_pair_metadata():
    response = client.get("/v1/config")
    assert response.status_code == 200
    json_data = response.json()
    assert json_data["service_type"] == "plain_baseline"
    assert "models" in json_data
    
    models = json_data["models"]
    assert "plain-llama-3-8b" in models
    assert "plain-llama-3.2-3b" in models
    assert "plain-qwen-0.5b" in models
    
    llama3 = models["plain-llama-3-8b"]
    assert llama3["hf_model_id"] == "meta-llama/Meta-Llama-3-8B-Instruct"
    assert llama3["paired_rif_model"] == "kalpana-llama-3-8b-rif"
    assert "context_limit" in llama3