AI service & Smart Routing
The AI service & Smart Routing page is where you manage which AI service Aureink uses and how it trades off between speed and capability. Most teams can use it out of the box; when you have special needs, you can also connect your own model service and fine-tune it.
1 · Overview
Every bit of Aureink’s thinking and execution is driven by an AI model service behind the scenes. This page lets you decide two things: whose service to use (the built-in one, or your own) and how routing trades off (faster responses, or stronger results). Different choices lead to different response speeds, output quality, and quota consumption.
2 · Built-in Aureink service
Aureink comes with a ready-to-use AI service, so most teams can use it directly without bringing their own API Key. You don’t need to worry about which model it connects to underneath — Aureink handles the routing for you.
3 · Smart Routing preferences
Smart Routing lets Aureink automatically trade off between “fast” and “strong” based on your preference. It offers three levels — just pick using the “Your scenario → Recommended” table below:
| Your scenario | Recommended level | Characteristics |
|---|---|---|
| Most tasks / not sure which to pick | Balanced | Balances speed and capability; the default choice |
| Everyday Q&A / want fast responses | Faster responses | Prioritizes response speed; good for simple, lightweight tasks |
| Complex reasoning / want high-quality output | Stronger results | Prioritizes capability; steadier on complex tasks, but slower and uses more quota |
Tip: When in doubt, use “Balanced”; if responses feel slow on a simple question, switch to “Faster responses,” and switch to “Stronger results” when you hit a complex task it can’t crack.
4 · Custom provider
If you need to connect your own model service, you can add a custom provider.
〔Figure: the custom provider configuration page, highlighting Type, API Key, Base URL, Model, and “Test connection”〕
- Fill in the basics — choose the Type and enter the API Key, Base URL, and Model.
- Fetch the model list — automatically pull the models the service supports, saving you from typing them by hand.
- Mark capabilities — tag the model with the capabilities it supports: Tool Use, Vision, and Extended Thinking, so Aureink can route correctly.
- Test connection — run a test before saving to confirm the credentials and address work.
- Import / export configuration — export the configuration as a backup, or import it to reuse on other devices.
Tip: A custom provider’s API Key is a sensitive credential — keep it safe and never leak it or commit it to a code repository; you usually obtain or refresh the Key in the console of your model service provider.
5 · Important notes
Important notes
- Security tip: keep your custom provider’s API Key safe; don’t send it out or write it into shared files.
- Quota consumption: different models have different capabilities and quota consumption — the “Stronger results” level and large models consume more, so prefer “Faster responses” or smaller models for simple tasks.
- Third-party sharing: when using a custom provider, task data is sent to the third-party service you configured, so confirm its data compliance before enabling it.
- Usage tip: after connecting, use “Test connection” or send a quick message to verify it responds correctly before putting it into real use.
Last updated · July 2026 ｜ Next: Permission modes →