Skip to content
aiqbench takeAI Products

aiqbench analysis

This article details a configuration-as-code approach for managing Large Language Model calls via OpenRouter presets, allowing developers to centralize system prompts, routing logic, and sampling parameters outside of application code. By treating LLM configurations like infrastructure modules, this method eliminates the need for redeployments when updating models or prompts across multiple applications. The guide focuses on practical implementation strategies for maintaining consistent AI behavior in diverse environments without hardcoding settings.
Source coveringOpenRouter
What this isA model's reading of the article and related coverage, written for aiqbench. Not the publisher's words.

How to Use OpenRouter Presets: Config-as-Code Guide

OpenRouter8 min read

A preset is config-as-code for your LLM calls: a named, versioned set of models, system prompts, provider routing, and sampling parameters. Define it once, reference it as @preset/your-slug everywhere, and update every app from the dashboard without a redeploy.

You’ve put the same model name, system prompt, and temperature in a web app, a batch script, and a notebook. Now you want to change the temperature or model, so you have to edit all three. We built presets to solve exactly this. A preset is config-as-code for your LLM calls, a named, versioned set of models, system prompts, provider routing, and sampling parameters. We let you define it once and use its name everywhere. When you change the preset, every app that uses it picks up the change. This guide adds to the presets docs reference and the launch announcement. You’ll create a preset, use it in an API call, and work through two setups: an image-prompt-enhancement preset and a fixed Fusion configuration. A preset stores a model or fallback array, a system prompt, provider routing, and sampling parameters under one name. Use it in any request as "model": "@preset/your-preset-name". Update it in the dashboard, and every app that…

Excerpt shown under the publisher's RSS terms · 160 words · full text and figures are on the publisher's site

Related coverage