# Models
URL: https://docs.valsight.ai/models/
Description: Submodels help structure large models by splitting logic into smaller, focused parts. They make complex models easier to understand, maintain, and work on to…
Submodels help structure large models by splitting logic into smaller, focused parts. They make complex models easier to understand, maintain, and work on together.

In Valsight, submodels are not isolated fragments. They remain fully connected and together form one coherent business model.

## When to use it

* When a model becomes large or hard to reason about
* When different parts of the logic serve different purposes

## Best practice: the Valsight “fish” structure

A recommended way to structure models is the **Valsight fish**, which typically spans **up to six levels**:

* **500\_Tail**: Data input models
* **400-20X\_Mid-sections**: Market logic, sales volumes, production costs, OPEX
* **100\_Head**: P\&L and KPI models

Each level represents a clear business responsibility and can be implemented as one or multiple submodels.
Submodels can be configured independently, depending on complexity and team needs.

This structure improves overview, supports scaling, and makes model logic easier to navigate.

![Valsight\_Fish.PNG](./images/6700335106.png)

## Node availability & model flow

Nodes are available for connections based on their position in the fish flow:

* Nodes can be used **within their own model**
* Nodes can be used in **any model that comes after them** in the flow

For example:

* Nodes from *Sales* are available in *P\&L* and *KPIs*
* Nodes from *KPIs* are **not** available in *Sales* or *P\&L*

To support tighter coordination, submodels can also have **two-way relationships** (e.g. *Sales* ↔ *Market Model*).
This allows nodes from both models to be used across the connected models.

## Access & permissions

Submodels allow **granular access control** across the model structure.

Permissions can be set so that a user:

* Has access only to specific submodels (e.g. OPEX)
* Automatically gains access to all preceding models in the flow
* Cannot access other models on the same level or any subsequent models

This makes it possible to safely collaborate across teams while protecting sensitive logic.

## FAQ

**Do submodels affect results?**

No. They only structure logic.

**Can I add submodels later?**

Yes. Submodels can be added at any time. What matters most is how responsibilities are separated, not how many submodels you create.

**Are submodels required?**

No. They are most useful for larger or collaborative models.

## Related documentation

* [Model Editor UI](/model-editor-ui/)