Model reference · open weights

TrOCR

TrOCR is an open-weight language model from qualcomm.

For self-hosting, compare this build's requirements below with the GPU you plan to rent.

• TrOCR is an image-to-text model developed by Qualcomm for optical character recognition, utilizing pre-trained image and text transformer architectures.
• The repository provides pre-exported model files optimized for Qualcomm devices, with the encoder containing 23.0M parameters and the decoder containing 38.3M parameters.
• The model uses a 320x320 input resolution and is available under the MIT license.

Summary of the qualcomm/TrOCR model card, 2026-10-01

Hugging Face ↗Paper ↗