NexoRouter

Documentation

LlamaIndex in the NexoRouter documentation.

LlamaIndex

Status: Candidate chat setup, not yet verified by NexoRouter.

LlamaIndex can use OpenAI-compatible chat models, but many LlamaIndex workflows also need embeddings. Keep chat and embeddings separate until NexoRouter publishes stable embeddings support.

Chat LLM example

import os
from llama_index.llms.openai import OpenAI

llm = OpenAI(
    api_key=os.environ["NEXOROUTER_API_KEY"],
    api_base="https://api.nexorouter.com/v1",
    model="deepseek-v4-flash",
)

response = llm.complete("Write one short production checklist.")
print(response)

Verification path

  1. Run one plain chat completion or completion-style call.
  2. Confirm the request appears in Usage Logs.
  3. Add indexes, retrievers, agents, or tools only after chat is proven.
  4. Use a separate verified embeddings provider if your index requires embeddings.

Boundaries

  • RAG indexes normally require embeddings.
  • Agent workflows depend on tool calling and model behavior.
  • Streaming and structured outputs need separate tests.
LlamaIndex — NexoRouter