# Train a Model

You can train computer vision models in the Roboflow interface.

Roboflow offers two training options:

* Roboflow Custom Train: Our flagship training service, ideal for creating production-ready models.
* [Neural Architecture Search](/train/neural-architecture-search.md): Discover novel model architectures and fine tune at the same time.

Also, when you approve your first batch of image annotations, a **Roboflow Instant Model** is automatically trained. These models can be used immediately for auto labeling or deployment.

Models trained on Roboflow can be deployed with Inference, our on-device inference server, or in the cloud using our Serverless Hosted API with Workflows, Batch Processing with Workflows, or with your model API endpoint.

{% hint style="info" %}
Read our [licensing guidance](https://roboflow.com/licensing) to learn more about how models trained on Roboflow are licensed.
{% endhint %}

### Train a Model

To train a computer vision model, visit "Train" in the project navigation.

<figure><img src="/files/hyP3UulOMlrDV2KxeH2L" alt=""><figcaption></figcaption></figure>

Click the "Custom Training" button to start configuring a training job:

<figure><img src="/files/QRvpvtfBDKPdqEFu5OOM" alt=""><figcaption></figcaption></figure>

#### Neural Architecture Search (NAS)

For object detection and instance segmentation projects, you can use **Neural Architecture Search (NAS)** instead of selecting a single model architecture. NAS automatically trains and evaluates multiple model configurations, then recommends the best one based on your accuracy and latency requirements.

To use NAS, select the NAS option when choosing a training engine. NAS requires at least 15 validation images.

For more details on NAS training charts, see [View Training Results](/train/training-results.md#nas-training-charts).

#### Select a Model Architecture

Next, you need to select a model architecture and size. This is the machine learning technology used to train your model.

The model architectures you can train depend on the type of project you have set up. Refer to the [Supported Models table](/deploy/supported-models.md) for details on training compatibility.

For object detection, RF-DETR offers the best accuracy. For instance segmentation, RF-DETR Seg (Preview) offers the best accuracy.

Model sizes will vary depending on the architecture of the model you choose. For example, RF-DETR — a state-of-the-art object detection model — offers Nano, Small, Medium, and Base. Medium, Large, and Extra Large are available only to paid users. Training a [SAM3](/deploy/supported-models/sam3.md) model is available on paid plans that include [usage-based billing](/billing/credits.md).

Choose an architecture available for your project type, then click "Continue":

<figure><img src="/files/v8DKAkRKUmOjT4tUTSQT" alt=""><figcaption></figcaption></figure>

#### Select a Checkpoint

After selecting a training option, you will be asked whether you want to train from a checkpoint. The tabs below show the configuration options for each model type.

{% tabs %}
{% tab title="Object Detection" %}
You have three options:

* **Train from a Previous Checkpoint:** Ideal for when you already have a working model that you want to improve.
* **Train from Public Checkpoint:** Ideal for your first model version, or for when a previous training run did not achieve the expected results.
* **Train from Random Initialization:** **For advanced users only**, this option gives you a blank slate from which to train. Most users see worse results when using this option.
  {% endtab %}

{% tab title="Classification/Semantic Segmentation" %}
For Classification and Semantic Segmentation models, only one checkpoint is available.
{% endtab %}
{% endtabs %}

<details>

<summary>How do I choose a training option?</summary>

We recommend training from a Public Checkpoint for new object detection projects. By default, we offer training from a model trained on the Microsoft COCO dataset. For classification and semantic segmentation, we only support training from an ImageNet.

You can train from checkpoints based on projects hosted on Universe (object detection only). To do so, first [star a project in Universe](https://blog.roboflow.com/launch-universe-model-checkpoint/). Then, the project will be available as a training checkpoint in the Roboflow web application.

Furthermore, you can train from a checkpoint based on a previous version of a model (object detection, instance segmentation, and keypoint detection only). This method allows for a faster training process. We only recommend training from a previous checkpoint for your model if your model achieves strong performance).

Training from a Checkpoint means that you are employing [Transfer Learning](https://blog.roboflow.com/what-is-transfer-learning/). Transfer Learning will initialize your model training from the model you have selected. This can help to reduce training time, and provide you with improved training scores.

Training from Scratch means that you are *not* employing Transfer Learning. This will initialize your model training with randomized initial values for the model weights.

</details>

#### Start the Training Job

Once you have chosen a Checkpoint from which to train, click Start Training.

Your dataset will then be zipped and prepared for training in the Roboflow cloud.

Before training starts, the training summary shows estimated duration and credit cost:

<figure><img src="/files/KbcmQkZzALvMYrP9DmzS" alt=""><figcaption></figcaption></figure>

The larger the dataset, and the larger the images in your dataset, the longer it will take for your model to train.

You will receive an email when the training process finishes. In most cases, this should be under 24 hours.

#### Pricing

Training on Roboflow is priced on the length of the train job. You can see more information on our [credits page](https://www.roboflow.com/credits).

If you are a student or researcher and need credits for a project on which you are working, you can [apply for additional credits](https://roboflow.com/contribute).


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.roboflow.com/train/train.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.