> ## Documentation Index
> Fetch the complete documentation index at: https://docs.tryblend.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Choose Output
> Pause the workflow and pick one or more upstream outputs to continue with.
The **Choose Output** node collects several candidate outputs, pauses the run, and lets you compare them side by side.
Use it to pick the one, or ones, that should continue downstream. For example, generate two images with different models, then decide which one moves forward.
## How it works
Connect multiple upstream outputs into the single **Choices** input. Each connected output becomes one candidate the node can show you.
When the run reaches this node, it **pauses and waits for a selection**. You see every candidate flattened into a list and pick from it.
Once you choose, the node forwards your selection through its **Selected** output and the workflow resumes.
Candidates must all be the **same content type** — all images, all text, all video, and so on. The node needs **at least two** comparable choices to be valid. While editing, the node shows a status badge:
* **awaiting selection** — choices are valid and ready to be picked from at run time.
* **needs matching choices** — choices are connected but don't all share one content type, or there aren't enough.
* **idle** — no choices connected yet.
## Inputs
| Handle | Type | Required | Notes |
| ------- | ----- | -------- | ------------------------------------------------------------------------------------------------------------------------ |
| Choices | `any` | Yes | Accepts multiple inputs. Connect two or more upstream outputs of the **same** content type (all images, all text, etc.). |
## Outputs
| Handle | Type | Notes |
| -------- | ----- | ------------------------------------------------------------------------------------------------------------------------------- |
| Selected | `any` | The candidate(s) you chose. In single mode this is one output item; in multiple mode it is one output item per selected choice. |
## Settings
| Setting | Options | What it does |
| -------------- | ---------------------------------------- | ----------------------------------------------------------------------------------------------------- |
| Selection mode | **Single choice** / **Multiple choices** | Single (default): forward exactly one candidate. Multiple: forward any number of selected candidates. |
Single choice collapses a list to one downstream item. Multiple choices preserves a list, so the next node runs once per selected item.
The configuration panel also shows a **Choices** preview listing the connected candidates and a hint such as how many will be presented, or why the current choices don't yet qualify.
Candidates that failed upstream can't be selected. You need at least one successful candidate for the node to continue.
## When to use
* **A/B comparing models** — run the same prompt through two image or video models and pick the better result.
* **Human-in-the-loop approval** — pause an automated pipeline so a person signs off on which output proceeds.
* **Curating multiple variations** — in multiple mode, keep several good options and forward them all together.
## Connects to
* **Upstream:** two or more generators producing the same kind of output — for example, several image model nodes, or multiple prompt/text nodes.
* **Downstream:** a [Result](/workflow/nodes/result) node, or any node that accepts the chosen content type.
## Next steps
How single and multiple choices affect downstream run counts.
How nodes, handles, and connections fit together.
Capture the chosen output as a final workflow result.