Plan required for new RFDETRLarge sizes

[Guy_Haiby]

I noticed that the version of fine-tuned rfdetr-large model I am using has been deprecated when I got this warning during inference: “WARNING: Automatically switched to deprecated model configuration, due to using deprecated weights. This will be removed in a future version.
Please retrain your model with the new weights and configuration.”

I am thinking this is because of the newer Large, XL, 2XL configurations referenced in this paper https://arxiv.org/pdf/2511.09554.

The (deprecated) rfdetr-large model that am currently using has 135M parameters, compared to 33M on the newer rfdetr-large. I am not sure how the resolution and patch size compare, but I noticed major improvement on classifying small objects when using the 135M large model over the 33M medium model.

My questions is if I wanted to keep the parameters consistent and use the XL or 2XL versions, would I need to upgrade from my current paid core plan to the enterprise plan?

For context I am using this model as part of a larger pipeline that is used to extract data from videos for commercial clients.

• Project Type: Object Detection
• Operating System & Browser: MacOS, Chrome
• Project Universe Link or Workspace/Project ID:
• Do you grant Roboflow Support permission to access your Workspace for troubleshooting? (Yes/No): Yes

[brad]

The RF-DETR XL and 2XL models (which use a larger backbone than the smaller sizes) are released under the Platform Model License which basically says “you can use them according to the terms of your Roboflow Platform plan.”

All Roboflow plans (including our free public, research, and startup plans) include the right to use these models in the Roboflow Cloud and self-hosted via Roboflow Inference. They will consume your credits just like any other model. This is the happy path and what we typically recommend.

The research plan additionally lets you use these models however you need to for non-commercial research purposes.

If you need to use the weights outside of our ecosystem (eg converting them for an embedded device we don’t yet support) that’s where you’ll need an enterprise plan and should talk to our sales team about your use-case.