RAG-система на Qwen3.8 Flash Next
с контекстом до 1M

import os
from langchain_openai import ChatOpenAI

llm = ChatOpenAI(
    base_url="https://api.42gpu.ru/vllm/v1",
    api_key=os.environ["GPU42_API_KEY"],
    model="qwen3.8-flash-next",
)

# Дальше — стандартный RAG chain
retriever = vectorstore.as_retriever()
chain = create_stuff_documents_chain(llm, prompt)
rag = create_retrieval_chain(retriever, chain)

import os
from llama_index.llms.openai_like import OpenAILike

llm = OpenAILike(
    api_base="https://api.42gpu.ru/vllm/v1",
    api_key=os.environ["GPU42_API_KEY"],
    model="qwen3.8-flash-next",
    context_window=1000000,
)

# Дальше — стандартный QueryEngine
index = VectorStoreIndex.from_documents(docs)
query_engine = index.as_query_engine(llm=llm)