> ## Documentation Index
> Fetch the complete documentation index at: https://docs.ideatura.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Models and credits

> Choose AI settings, understand usage and compare task costs.

<h2 id="models-and-thinking">
  1\. Models and thinking
</h2>

<span id="models-and-thinking-overview" />

Use the model picker beside the prompt to select an available model. Where the model offers several reasoning levels, select the thinking level appropriate to the task.

Model, thinking level, Ideatura Mode, and Connected Apps access are different controls. Increasing one does not select the correct document or fix incomplete references.

<span id="models-and-thinking-in-ideatura" />

1. Open the model picker and choose from the account’s available models.
2. Check the thinking level where shown.
3. Use a small representative task to compare output quality, corrections, time, and credits.
4. Check the selected model before continuing a task after a model change.

<span id="models-and-thinking-check-the-result" />

The selected model and thinking level are visible, and the task has enough context to produce an inspectable result.

<h2 id="credits-and-storage">
  2\. Credits and storage
</h2>

<span id="credits-and-storage-overview" />

Open Settings and look at Subscription. Read the Monthly Credits and Daily Credits allowances shown for your account, and check Workspace Storage separately.

Record usage before and after a bounded task when assessing its cost. Include retries and corrections, and avoid attributing a balance change to one task if other work ran at the same time.

<span id="credits-and-storage-in-ideatura" />

1. Open Settings → Subscription.
2. Read the applicable monthly and daily usage displays.
3. Record the baseline before the example.
4. Run and check the task, then refresh and compare usage.

<span id="credits-and-storage-check-the-result" />

The example explains credits consumed to reach a checked result, without confusing storage with credits.

<h2 id="reduce-rework">
  3\. Reduce rework
</h2>

<span id="reduce-rework-overview" />

For a repeated drawing task, establish the rule once, test it on one view, and reuse the checked automation with appropriate parameters. For a new modelling task, provide the drawings, dimensions, and missing assumptions before requesting a large reconstruction.

Use small loops: inspect, plan, run, review, and correct one issue. Compare the total credits and review effort needed for a usable result, including unsuccessful attempts.

<span id="reduce-rework-in-ideatura" />

1. Choose a representative small scope.
2. Provide the exact view, rules and references.
3. Check the result before expanding the scope or rerunning.
4. Record total usage and correction work.

<span id="reduce-rework-check-the-result" />

The comparison ends at an accepted model or drawing, rather than at the first response.

<h2 id="tokens-and-credits">
  4\. Tokens and credits
</h2>

<span id="tokens-and-credits-overview" />

Context measures what can fit into a model request. Token usage measures processed input and output across requests. Credits are Ideatura’s usage unit; Workspace Storage measures retained files.

Read the usage indicator’s tooltip for its label and detail. Compare conversation usage with the account allowance in Settings rather than treating the two figures as interchangeable.

<span id="tokens-and-credits-in-ideatura" />

1. Read the credit tooltip for detail.
2. Compare the conversation total with account meters in General.
3. Keep token or context indicators labelled by their actual unit.
4. Record before/after values for a bounded task.

<span id="tokens-and-credits-check-the-result" />

You do not mistake remaining context for remaining credits, or a rounded zero for free usage.

<h2 id="usage-factors">
  5\. Usage factors
</h2>

<span id="usage-factors-overview" />

Model choice, reasoning effort, the amount of context processed, generated output and repeated tool/repair cycles can affect the amount of AI work in a task. A short final reply does not establish that the task was inexpensive.

Check your account’s charging terms when estimating a task. Do not assume one token equals one credit, or that attachments, retries and saved automations carry a fixed charge.

<span id="usage-factors-in-ideatura" />

1. Keep task scope and input constant when comparing options.
2. Record model, thinking level and agent route.
3. Measure the observed usage and inspect output quality.
4. Choose the smallest setting that reliably meets the task.

<span id="usage-factors-check-the-result" />

A model or setting is judged by useful checked output and observed cost.

<h2 id="automation-costs">
  6\. Automation costs
</h2>

<span id="automation-costs-overview" />

Creating or adapting an automation and running an already saved automation are different operations. Measure them separately before making a cost comparison.

Use a sample run, note the conversation and account usage before and after, and check the native result. A saved workflow is not automatically free, and a long conversation’s total is not the price of its most recent run.

<span id="automation-costs-in-ideatura" />

1. Record initial usage, route, model and task scope.
2. Create or adapt the automation and record the change.
3. Measure a separate saved-automation run.
4. Compare cost alongside errors, corrections and useful output.

<span id="automation-costs-check-the-result" />

You can explain the measured cost of authoring and execution separately.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.