Skip to main content

​
Overview

Haystack’s OpenAIChatGenerator supports custom OpenAI-compatible deployments through api_base_url. Point it at TokenLab and provide a TokenLab API key.
Type: RAG and pipeline frameworkPrimary Path: OpenAI-compatible Chat CompletionsSupport Confidence: Supported OpenAI-compatible path

​
Environment

export TOKENLAB_API_KEY="sk-your-tokenlab-key"

​
Generator

from haystack.components.generators.chat import OpenAIChatGenerator
from haystack.dataclasses import ChatMessage
from haystack.utils import Secret

generator = OpenAIChatGenerator(
    api_key=Secret.from_env_var("TOKENLAB_API_KEY"),
    api_base_url="https://api.tokenlab.sh/v1",
    model="claude-sonnet-5",
)

response = generator.run(
    messages=[ChatMessage.from_user("Summarize this document in one sentence.")]
)

​
Endpoint Notes

Use this path for Haystack chat generator pipelines. For embeddings, configure the matching Haystack embedding component with TokenLab’s /v1 base URL if your pipeline uses TokenLab embeddings.