Choosing a model

The default is a real choice, not a fallback

If you say nothing, an AI tool uses the model we have configured for it. That is deliberately chosen for value rather than for headline quality, and for most work it is the right answer.

Tiers

Models are grouped into three tiers:

Asking for a tier rather than a specific model is usually the better choice, because a named model eventually gets retired and your code keeps working if you asked for a tier.

Over the API

Send a model parameter with a tier name, a model id, or a model name:

Bash
curl -X POST "https://api.toolsxpo.com/v1/text-summarizer" \
  -H "Authorization: Bearer $TOOLSXPO_KEY" \
  -H "Content-Type: application/json" \
  -d '{"text":"…","model":"economy"}'

In the dashboard

Dashboard → API configuration lets you set a default model per tool, for the whole workspace or for one API key — so a production key and a test key can use different models without changing any code.

If your plan does not include it

Some models are limited to paid plans. If you ask for one you cannot use, the call still runs on the tool's default model rather than failing. We would rather return a working answer than a 403 you have to write code around.

The response always reports which model actually answered, so nothing is hidden — and because a cheaper model sits on a cheaper tier, running on it simply costs you less. There is no situation where a fallback charges you more than the model you asked for.

Your own instructions

Alongside the model you can save your own instructions — tone, format, house style. They are added on top of the tool's own instructions, not in place of them, so the tool still does its job the way it is documented to.

They are charged as input tokens on every call.