Configs

A Config represents a named configuration endpoint that can be referenced in API calls. Think of it as a specific LLM use case you want to manage, like "Customer Support Bot" or "Code Review Assistant".

Overview

Each config:

• Has a unique name and auto-generated config id.
• Can have multiple variants attached
• Is referenced in API calls via the x-llmops-config header

Creating a Config

Navigate to the Configs section in the LLMOps dashboard to create a new config.

Each config has:

Using Configs in API Calls

Reference your config using the x-llmops-config header:

curl -X POST http://localhost:3000/llmops/api/genai/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "x-llmops-config: xK9mP2nQ" \
  -d '{"messages": [{"role": "user", "content": "Hello!"}]}'

You can use either the config's UUID or its short slug.

With the OpenAI SDK

import OpenAI from 'openai';

const openai = new OpenAI({
  baseURL: 'http://localhost:3000/llmops/api/genai',
  apiKey: '', // Not needed for same-origin requests
  defaultHeaders: {
    'x-llmops-config': 'xK9mP2nQ',
  },
});

const response = await openai.chat.completions.create({
  model: 'gpt-4o-mini', // Can be overridden by variant
  messages: [{ role: 'user', content: 'Hello!' }],
});

Config Variants

A config can have multiple variants attached. Each variant represents a different model configuration (provider, model, parameters, system prompt).

Different environments can serve different variants of the same config:

This allows you to test new model configurations safely before promoting them to production.

Example Use Cases

Each of these could have multiple variants with different models, parameters, or system prompts depending on the environment.