Supported Models

All models you can deploy with Roboflow.

With Roboflow, you can deploy object detection, segmentation, classification, keypoint, and multimodal models.

These models can be deployed with Workflows using:

• Serverless Hosted API

• Dedicated Deployment (CPU or GPU)

• Self-hosted (Roboflow Inference)

The table below shows common models and their support for training, model weights upload and model weights export/download.

The table describes models that can be deployed with the Serverless Hosted API (including V1 and V2).

If a model is not supported on a Serverless Hosted API, it must be deployed either with a Dedicated Deployment with a GPU or on your own hardware with Roboflow Inference.

All models can be used with Batch Processing.

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

Model	Task Type(s)	Training Supported on Roboflow?	Model Weights Upload Supported?	Model Weights Export Supported?	Supported on Serverless Hosted API
Roboflow 3.0	Object Detection, Classification, Keypoint Detection, Instance Segmentation, Semantic Segmentation
RF-DETR	Object Detection, Instance Segmentation
YOLO26	Object Detection, Keypoint Detection, Instance Segmentation
YOLOv12	Object Detection
YOLO11	Object Detection, Instance Segmentation, Keypoint Detection
YOLOv9	Object Detection
YOLOv8	Object Detection, Instance Segmentation, Keypoint Detection
YOLOv8	Classification
YOLOv7	Instance Segmentation
YOLOv5	Object Detection, Instance Segmentation
Roboflow Instant	Object Detection
ViT	Classification
ResNet	Classification
Florence 2	Multimodal
PaliGemma 2	Multimodal, Object Detection
Qwen2.5-VL	Multimodal
SmolVLM2	Multimodal
SmolVLM 256M	Multimodal

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.