> For the complete documentation index, see [llms.txt](https://docs.roboflow.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.roboflow.com/models/supported-models.md).

# Supported Models

With Roboflow, you can deploy object detection, segmentation, classification, keypoint, and multimodal models. Models you have trained are referenced by a [model ID](/models/model-ids.md).

These models can be deployed with [Workflows](https://docs.roboflow.com/workflows) using:

* [Serverless Cloud API](https://docs.roboflow.com/deployment/roboflow-cloud/serverless-api)
* [Dedicated Deployment (CPU or GPU)](https://docs.roboflow.com/deployment/roboflow-cloud/dedicated-deployments)
* Self-hosted ([Roboflow Inference](https://docs.roboflow.com/deployment/self-hosted/self-hosted))

The table below shows common models and their support for training, [model weights upload](/models/model-weights/upload-custom-weights.md) and [model weights export/download](/models/model-weights/download-roboflow-model-weights.md), and Serverless Cloud API deployment.

The table describes models that can be deployed with the Serverless Cloud API.

{% hint style="info" %}
If a model is not supported on a Serverless Cloud API, it must be deployed either with a [Dedicated Deployment](https://docs.roboflow.com/deployment/roboflow-cloud/dedicated-deployments) with a GPU or on your own hardware with Roboflow Inference.
{% endhint %}

All models can be used with [Batch Processing](https://docs.roboflow.com/deployment/roboflow-cloud/batch-processing).

We recommend using Dedicated Deployments or Batch Processing with a GPU configured for optimal performance when using multimodal models.

### Object Detection

<table data-full-width="true" data-search="false"><thead><tr><th>Model</th><th width="133.8125">Task Type(s)</th><th data-type="checkbox">Training Supported on Roboflow?</th><th data-type="checkbox">Model Weights Upload Supported?</th><th data-type="checkbox">Model Weights Export Supported?</th><th data-type="checkbox">Supported on Serverless Cloud API</th></tr></thead><tbody><tr><td><a href="/pages/mKCYCvydcuCjknyGQxCu">RF-DETR</a></td><td>Object Detection</td><td>true</td><td>true</td><td>true</td><td>true</td></tr><tr><td><a href="/pages/orjCYkIgrDBK1bUw8YR3">Roboflow 3.0</a></td><td>Object Detection</td><td>true</td><td>true</td><td>true</td><td>true</td></tr><tr><td><a href="/pages/WQbqR8mx6H8YMNn2VraW">YOLOlite</a></td><td>Object Detection</td><td>true</td><td>false</td><td>true</td><td>true</td></tr><tr><td><a href="/pages/QNKRADmwtWRwnmrWdI8U">YOLO26</a></td><td>Object Detection</td><td>true</td><td>true</td><td>true</td><td>true</td></tr><tr><td><a href="/pages/DMf5zHlWHkAyX5UxzVLG">Roboflow Instant</a></td><td>Object Detection</td><td>true</td><td>false</td><td>true</td><td>true</td></tr><tr><td><a href="/pages/2Ze2rgmO0Nu3Zs3RCUXT">YOLOv12</a></td><td>Object Detection</td><td>true</td><td>true</td><td>true</td><td>true</td></tr><tr><td><a href="/pages/9yc9deacLKrdMTZQsIjl">PaliGemma 2</a></td><td>Object Detection</td><td>true</td><td>true</td><td>true</td><td>true</td></tr><tr><td><a href="/pages/Y65a7XwLS4uOifpbMMAo">YOLO11</a></td><td>Object Detection</td><td>true</td><td>true</td><td>true</td><td>true</td></tr><tr><td><a href="/pages/OqOqE2jdGKShiR0iB3OJ">YOLOv9</a></td><td>Object Detection</td><td>false</td><td>true</td><td>true</td><td>true</td></tr><tr><td><a href="/pages/kuNYAX1Ihkgmp5jzKf06">YOLOv5</a></td><td>Object Detection</td><td>false</td><td>true</td><td>true</td><td>true</td></tr><tr><td><a href="/pages/ZqETiFdXMKwrwn0zsw7z">OWLv2</a></td><td>Object Detection</td><td>false</td><td>false</td><td>false</td><td>false</td></tr><tr><td><a href="/pages/rQoYNRkthbFuOCvcpPin">Grounding DINO</a></td><td>Object Detection</td><td>false</td><td>false</td><td>false</td><td>false</td></tr></tbody></table>

### Instance Segmentation

<table data-full-width="true" data-search="false"><thead><tr><th>Model</th><th width="139.1875">Task Type(s)</th><th data-type="checkbox">Training Supported on Roboflow?</th><th data-type="checkbox">Model Weights Upload Supported?</th><th data-type="checkbox">Model Weights Export Supported?</th><th data-type="checkbox">Supported on Serverless Cloud API</th></tr></thead><tbody><tr><td><a href="/pages/mKCYCvydcuCjknyGQxCu">RF-DETR</a></td><td>Instance Segmentation</td><td>true</td><td>true</td><td>true</td><td>true</td></tr><tr><td><a href="/pages/orjCYkIgrDBK1bUw8YR3">Roboflow 3.0</a></td><td>Instance Segmentation</td><td>true</td><td>true</td><td>true</td><td>true</td></tr><tr><td><a href="/pages/0Ta3KtKxAoafBjFmVqgm">SAM3</a></td><td>Instance Segmentation</td><td>true</td><td>false</td><td>false</td><td>true</td></tr><tr><td><a href="/pages/QNKRADmwtWRwnmrWdI8U">YOLO26</a></td><td>Instance Segmentation</td><td>true</td><td>true</td><td>true</td><td>true</td></tr><tr><td><a href="/pages/Y65a7XwLS4uOifpbMMAo">YOLO11</a></td><td>Instance Segmentation</td><td>true</td><td>true</td><td>true</td><td>true</td></tr><tr><td><a href="/pages/RxWrBEXnEHQrfLs9l7ph">SAM2</a></td><td>Instance Segmentation</td><td>false</td><td>false</td><td>false</td><td>true</td></tr><tr><td><a href="/pages/YQHFd6SnuHXHuOsvVkoH">YOLOv7</a></td><td>Instance Segmentation</td><td>false</td><td>true</td><td>true</td><td>true</td></tr></tbody></table>

### Classification

<table data-full-width="true" data-search="false"><thead><tr><th>Model</th><th width="138.1015625">Task Type(s)</th><th data-type="checkbox">Training Supported on Roboflow?</th><th width="165.47265625" data-type="checkbox">Model Weights Upload Supported?</th><th data-type="checkbox">Model Weights Export Supported?</th><th data-type="checkbox">Supported on Serverless Cloud API</th></tr></thead><tbody><tr><td><a href="/pages/orjCYkIgrDBK1bUw8YR3">Roboflow 3.0</a></td><td>Classification</td><td>true</td><td>false</td><td>false</td><td>true</td></tr><tr><td><a href="/pages/P9mRXPQw4MQuYD7yBLa1">DINOv3</a></td><td>Classification</td><td>true</td><td>false</td><td>false</td><td>true</td></tr><tr><td><a href="/pages/QhpCBxwXabfkKxIlSWo8">ViT</a></td><td>Classification</td><td>true</td><td>true</td><td>false</td><td>true</td></tr><tr><td><a href="/pages/ptvJbudgHzl6hmrmQlnX">ResNet</a></td><td>Classification</td><td>true</td><td>true</td><td>false</td><td>true</td></tr></tbody></table>

### Keypoint Detection

<table data-full-width="true" data-search="false"><thead><tr><th>Model</th><th width="134.7578125">Task Type(s)</th><th data-type="checkbox">Training Supported on Roboflow?</th><th data-type="checkbox">Model Weights Upload Supported?</th><th data-type="checkbox">Model Weights Export Supported?</th><th data-type="checkbox">Supported on Serverless Cloud API</th></tr></thead><tbody><tr><td><a href="/pages/mKCYCvydcuCjknyGQxCu">RF-DETR (Preview)</a></td><td>Keypoint Detection</td><td>true</td><td>false</td><td>false</td><td>false</td></tr><tr><td><a href="/pages/orjCYkIgrDBK1bUw8YR3">Roboflow 3.0</a></td><td>Keypoint Detection</td><td>true</td><td>true</td><td>true</td><td>true</td></tr><tr><td><a href="/pages/QNKRADmwtWRwnmrWdI8U">YOLO26</a></td><td>Keypoint Detection</td><td>true</td><td>true</td><td>true</td><td>true</td></tr><tr><td><a href="/pages/Y65a7XwLS4uOifpbMMAo">YOLO11</a></td><td>Keypoint Detection</td><td>true</td><td>true</td><td>true</td><td>true</td></tr></tbody></table>

### Semantic Segmentation

<table data-full-width="true" data-search="false"><thead><tr><th>Model</th><th width="133.90625">Task Type(s)</th><th data-type="checkbox">Training Supported on Roboflow?</th><th data-type="checkbox">Model Weights Upload Supported?</th><th data-type="checkbox">Model Weights Export Supported?</th><th data-type="checkbox">Supported on Serverless Cloud API</th></tr></thead><tbody><tr><td><a href="/pages/QNKRADmwtWRwnmrWdI8U">YOLO26</a></td><td>Semantic Segmentation</td><td>true</td><td>true</td><td>true</td><td>true</td></tr><tr><td><a href="/pages/uriJHIFC98EbNx02BDzG">Roboflow 2.0</a></td><td>Semantic Segmentation</td><td>true</td><td>false</td><td>false</td><td>true</td></tr></tbody></table>

### Multimodal

<table data-full-width="true" data-search="false"><thead><tr><th>Model</th><th width="133.90625">Task Type(s)</th><th data-type="checkbox">Training Supported on Roboflow?</th><th data-type="checkbox">Model Weights Upload Supported?</th><th data-type="checkbox">Model Weights Export Supported?</th><th data-type="checkbox">Supported on Serverless Cloud API</th></tr></thead><tbody><tr><td><a href="/pages/5rOs3EhPQay52zr41RiY">Cosmos 3 Edge</a></td><td>Multimodal</td><td>false</td><td>false</td><td>false</td><td>true</td></tr><tr><td><a href="/pages/sGRiR4sHpVy5zTUusEkK">Qwen3.5</a></td><td>Multimodal</td><td>true</td><td>true</td><td>false</td><td>false</td></tr><tr><td><a href="/pages/ebfLFlCv8vu6xvjVSkjF">Qwen3-VL</a></td><td>Multimodal</td><td>true</td><td>true</td><td>false</td><td>true</td></tr><tr><td><a href="/pages/QZbBNLWHZPOdvxRe9AyF">Perception Encoder</a></td><td>Multimodal</td><td>false</td><td>false</td><td>false</td><td>false</td></tr><tr><td><a href="/pages/EiY2lYYOEZyPo5l1dyDh">SmolVLM2</a></td><td>Multimodal</td><td>true</td><td>false</td><td>false</td><td>false</td></tr><tr><td>Qwen2.5-VL</td><td>Multimodal</td><td>true</td><td>true</td><td>false</td><td>false</td></tr><tr><td>SmolVLM 256M</td><td>Multimodal</td><td>true</td><td>false</td><td>false</td><td>false</td></tr><tr><td><a href="/pages/9yc9deacLKrdMTZQsIjl">PaliGemma 2</a></td><td>Multimodal</td><td>true</td><td>true</td><td>true</td><td>true</td></tr><tr><td><a href="/pages/Uc5ItrvOWrXZ41tUy8iC">Florence 2</a></td><td>Multimodal</td><td>true</td><td>true</td><td>false</td><td>true</td></tr><tr><td><a href="/pages/Zbo1sQWU6HdhY8UQk4aj">Moondream2</a></td><td>Multimodal</td><td>false</td><td>false</td><td>false</td><td>false</td></tr><tr><td><a href="/pages/2MgFtbn5ltbZHVYRazNg">YOLO-World</a></td><td>Multimodal</td><td>false</td><td>false</td><td>false</td><td>true</td></tr><tr><td><a href="/pages/OKZLC1fAim75ro3hH44U">CLIP</a></td><td>Multimodal</td><td>false</td><td>false</td><td>false</td><td>true</td></tr></tbody></table>

### OCR

<table data-full-width="true" data-search="false"><thead><tr><th>Model</th><th width="133.90625">Task Type(s)</th><th data-type="checkbox">Training Supported on Roboflow?</th><th data-type="checkbox">Model Weights Upload Supported?</th><th data-type="checkbox">Model Weights Export Supported?</th><th data-type="checkbox">Supported on Serverless Cloud API</th></tr></thead><tbody><tr><td><a href="/pages/vCZZlimnh5rIEFbmLUjX">GLM-OCR</a></td><td>OCR</td><td>false</td><td>false</td><td>false</td><td>true</td></tr><tr><td><a href="/pages/xVONbFZDCSLAjOcUSzhk">TrOCR</a></td><td>OCR</td><td>false</td><td>false</td><td>false</td><td>false</td></tr><tr><td><a href="/pages/UpqCnNDConBnJCPh7LQE">DocTR</a></td><td>OCR</td><td>false</td><td>false</td><td>false</td><td>true</td></tr><tr><td><a href="/pages/pEl5Wti6iVuUrvGnmwSH">EasyOCR</a></td><td>OCR</td><td>false</td><td>false</td><td>false</td><td>true</td></tr></tbody></table>

### Other

<table data-full-width="true" data-search="false"><thead><tr><th>Model</th><th width="133.90625">Task Type(s)</th><th data-type="checkbox">Training Supported on Roboflow?</th><th data-type="checkbox">Model Weights Upload Supported?</th><th data-type="checkbox">Model Weights Export Supported?</th><th data-type="checkbox">Supported on Serverless Cloud API</th></tr></thead><tbody><tr><td><a href="/pages/9RL9eZJD5v3qkVfjTImJ">Depth Anything V2 Small</a></td><td>Depth Estimation</td><td>false</td><td>false</td><td>false</td><td>false</td></tr><tr><td><a href="/pages/tncFqcswFUY3wZigHM11">L2CS-Net</a></td><td>Gaze Detection</td><td>false</td><td>false</td><td>false</td><td>true</td></tr></tbody></table>

### Models Supported in Workflows

You can run all of the above models in Workflows, as well as other models like Segment Anything 2, CLIP, OpenAI's GPT models, and more. [See a full list of models you can run in a Workflow.](https://docs.roboflow.com/workflows/blocks/blocks#Models)

## Foundation Models

Foundation models are trained on very large, general datasets, so they work out of the box without you training anything. OpenAI's CLIP, for example, was trained on over 400 million image-text pairs, which lets it classify and compare images across a wide range of domains.

Roboflow supports foundation models for segmentation, zero-shot detection, embeddings, OCR, depth estimation, and visual question answering. Every model listed on this page is callable over HTTP, so you do not have to set up or configure each one yourself.

Common uses:

* **Zero-shot detection and segmentation.** [SAM3](/models/supported-models/sam3.md), [Grounding DINO](/models/supported-models/grounding-dino.md), [YOLO-World](/models/supported-models/yolo-world.md), and [OWLv2](/models/supported-models/owlv2.md) find objects from a text prompt or a single example, with no training data.
* **Automated labeling.** Use a zero-shot model to label a dataset, then train a smaller, faster model that is fine-tuned to your use case.
* **Embeddings.** [CLIP](/models/supported-models/clip.md) and [Perception Encoder](/models/supported-models/perception-encoder.md) turn images and text into vectors for clustering, deduplication, content moderation, and dataset search.
* **Reading and describing images.** [Florence 2](/models/supported-models/florence-2.md), [Qwen3-VL](/models/supported-models/qwen3-vl.md), [Moondream2](/models/supported-models/moondream2.md), and the [OCR models](/models/supported-models/doctr.md) answer questions about images and extract text.

## Pretrained Models

Roboflow ships COCO-pretrained checkpoints under short aliases such as `rfdetr-small` and `yolov8n-640`. You can pass an alias anywhere a model ID is accepted, with no training and no project setup. See [Pretrained Model Aliases](/models/pretrained-aliases.md) for the full list.

## Running Your Own Models

To run a model you trained (or [uploaded weights](/models/model-weights/upload-custom-weights.md) for), you need its [model ID](/models/model-ids.md). Open your [Roboflow dashboard](https://app.roboflow.com), select the model, and click "Deploy" in the sidebar to find the ID, which has the form `{workspace}/{project}/{version}`.

Then pass it as `model_id`:

{% tabs %}
{% tab title="SDK (Python)" icon="python" %}

```python
import os
from inference_sdk import InferenceHTTPClient

client = InferenceHTTPClient(
    api_url="https://serverless.roboflow.com",
    api_key=os.environ["ROBOFLOW_API_KEY"],
)

results = client.infer("image.jpeg", model_id="taylor-swift-records/3")
print(results["predictions"])
```

{% endtab %}

{% tab title="Inference (native)" icon="server" %}

```python
from inference import get_model

model = get_model(model_id="taylor-swift-records/3")

results = model.infer("image.jpeg")[0]
print(results.predictions)
```

{% endtab %}
{% endtabs %}

Load the results into [supervision](https://supervision.roboflow.com/) with `sv.Detections.from_inference(results)` to visualize them.

## Community Models

[Roboflow Universe](https://universe.roboflow.com) hosts more than 50,000 models published by other Roboflow users, covering everything from defect detection to sports analytics and wildlife identification. Universe model IDs work exactly like your own: see [Find a Model on Universe](https://docs.roboflow.com/datasets/universe/universe/find-a-model-on-universe).

## Licensing

Models supported by Roboflow carry the licenses of their upstream projects. See [Roboflow Licensing](https://roboflow.com/licensing) for how Roboflow Inference and the models it serves are licensed.