PaddleOCR-VL 1.5

PaddleOCR-VL 1.5 needs an estimated 1.2 GB of VRAM at Q4, 1.7 GB at Q8 and 2.7 GB at FP16, based on its 959M parameters. All VRAM figures are estimates: weights plus about 15% and 0.5 GB for a short context.

Yes. A 24 GB, 16 GB, 12 GB or 8 GB GPU fits PaddleOCR-VL 1.5 at FP16 with an estimated 2.7 GB of VRAM.

Yes. PaddleOCR-VL 1.5 needs an estimated 1.2 GB at Q4, and assuming macOS lets the GPU use about 75% of unified memory, the smallest common Mac memory size that fits the Q4 estimate is 8 GB.

Yes. PaddleOCR-VL 1.5 has open weights, so you can self-host it for full control of your data.

Run PaddleOCR-VL 1.5 locally

PaddleOCR-VL 1.5 accepts text and image inputs. Audio, video and PDF are not listed among its input modalities.

The closest alternatives, matched on model kind and release date with no comparable price, are DeepSeek OCR 2 by DeepSeek, Trinity Large Preview by Arcee AI and Nemotron Content Safety Reasoning 4B by NVIDIA.

DeepSeek OCR 2PaddleOCR-VL 1.5 vs DeepSeek OCR 2Trinity Large PreviewPaddleOCR-VL 1.5 vs Trinity Large PreviewNemotron Content Safety Reasoning 4BPaddleOCR-VL 1.5 vs Nemotron Content Safety Reasoning 4B

PaddleOCR-VL 1.5 is open weights, released under the Apache 2.0 license.

Yes. PaddleOCR-VL 1.5 uses the Apache 2.0 license, a permissive license that allows commercial use.

The PaddleOCR-VL 1.5 weights are available in the Hugging Face repository PaddlePaddle/PaddleOCR-VL-1.5.

Run PaddleOCR-VL 1.5 locally

PaddleOCR-VL 1.5 was made by Baidu and released on January 29, 2026.

More from Baidu

Yes. PaddleOCR-VL 1.5 is offered as open weights.