Qianfan-OCR
Baidu · baidu/Qianfan-OCR
Weights license: Apache 2.0 · code and components may use other terms
Quality
No comparable score
We track no public, comparable benchmark that covers this model yet. No score is inferred.
Reported by the publisher
- OmniDocBench v1.5 (Overall Score)
- 93.12Document Parsing
- OlmOCR Bench (Overall Score)
- 79.8OCR
- OCRBench (Score)
- 880OCR
- Overall ↑
- 93.12Qianfan-OCR (Ours) · OmniDocBench v1.5 (Document Parsing) · End-to-end
- TextEdit ↓
- 0.041Qianfan-OCR (Ours) · OmniDocBench v1.5 (Document Parsing) · End-to-end
- FormulaCDM ↑
- 92.43Qianfan-OCR (Ours) · OmniDocBench v1.5 (Document Parsing) · End-to-end
- TableTEDs ↑
- 91.02Qianfan-OCR (Ours) · OmniDocBench v1.5 (Document Parsing) · End-to-end
- TableTEDss ↑
- 93.85Qianfan-OCR (Ours) · OmniDocBench v1.5 (Document Parsing) · End-to-end
Every reported result
- R-orderEdit ↓
- 0.049Qianfan-OCR (Ours) · OmniDocBench v1.5 (Document Parsing) · End-to-end
- OCRBench
- 880Qianfan-OCR (Ours) · General OCR Benchmarks
- CCOCR-multilan
- 76.7Qianfan-OCR (Ours) · General OCR Benchmarks
- CCOCR-overall
- 79.3Qianfan-OCR (Ours) · General OCR Benchmarks
- DocVQA
- 92.8Qianfan-OCR · Document Understanding
- CharXiv_DQ
- 94.0Qianfan-OCR · Document Understanding
- CharXiv_RQ
- 85.2Qianfan-OCR · Document Understanding
- ChartQA
- 88.1Qianfan-OCR · Document Understanding
- ChartQAPro
- 42.9Qianfan-OCR · Document Understanding
- ChartBench
- 85.9Qianfan-OCR · Document Understanding
- TextVQA
- 80.0Qianfan-OCR · Document Understanding
- OCRVQA
- 66.8Qianfan-OCR · Document Understanding
- Overall
- 87.9Qianfan-OCR (Ours) · Key Information Extraction (KIE)
- OCRBench KIE
- 95.0Qianfan-OCR (Ours) · Key Information Extraction (KIE)
- OCRBenchv2 KIE (en)
- 82.8Qianfan-OCR (Ours) · Key Information Extraction (KIE)
- OCRBenchv2 KIE (zh)
- 82.3Qianfan-OCR (Ours) · Key Information Extraction (KIE)
- CCOCR KIE
- 92.8Qianfan-OCR (Ours) · Key Information Extraction (KIE)
- Nanonets KIE (F1)
- 86.5Qianfan-OCR (Ours) · Key Information Extraction (KIE)
Values as printed on the card, with the publisher's own settings (card section: Benchmark Results › OmniDocBench v1.5 (Document Parsing); Benchmark Results › General OCR Benchmarks). Not comparable across publishers and not used in any ranking. Model card
Runs on
Fast memory holds it 118
The planning estimate fits the memory the model can use quickly: graphics card VRAM, or unified memory within the macOS GPU share.
Show all 118
Within memory, outside the fast part 1
Within the machine's memory, outside its fast part: on a graphics card the layers that do not fit run on the CPU from PC RAM (the listed RAM, or 32 GB assumed); on a Mac, above the macOS default GPU share. Not a speed figure.
Hardware profiles grouped by the Q4 planning estimate on each machine (dedicated GPUs reserve 2 GB; unified and system memory 25%, 2 to 8 GB): the fast memory holds it, it is within memory but outside the fast part (layers offloaded to PC RAM, or above the macOS GPU share), or the machine is one memory tier short. Not a measured fit or speed.
Baidu released Qianfan-OCR on 2026-03-18: a vision-language model that answers about images, with 4.74B parameters on the Hugging Face count and the Apache 2.0 license. Languages the publisher declares: Multilingual.
License: commercial use allowed. At Q4 the weights load about 3.6 GB before each machine's reserve; image resolution, video frames and context add memory, and the vision encoder needs support in the runtime you pick.
Measured speed
No published measurement for this checkpoint
Tokens per second are shown only when a source publishes them with hardware, build and setup. They are never estimated.
How to run
Runtime compatibility and tokens per second have not been measured for this entry. Published metadata and a memory estimate do not prove a successful local run.
Library in source: transformers
Documented ways to run it
Copied as printed in the linked source (read 2026-10-18); not run by this site.
from transformers import AutoModelForImageTextToText, AutoProcessor
import torch
from PIL import Image
MODEL_PATH = "baidu/Qianfan-OCR"
model = AutoModelForImageTextToText.from_pretrained(
MODEL_PATH,
torch_dtype=torch.bfloat16,
device_map="auto",
).eval()
processor = AutoProcessor.from_pretrained(MODEL_PATH)
image = Image.open("./examples/document.png").convert("RGB")
prompt = "Parse this document to Markdown."
messages = [
{
"role": "user",
"content": [
{"type": "image", "image": image},
{"type": "text", "text": prompt},
],
},
]
inputs = processor.apply_chat_template(
messages,
add_generation_prompt=True,
tokenize=True,
return_dict=True,
return_tensors="pt",
).to(model.device)
with torch.no_grad():
output_ids = model.generate(
**inputs,
max_new_tokens=512,
do_sample=False,
)
generated_ids = output_ids[:, inputs["input_ids"].shape[1]:]
response = processor.batch_decode(generated_ids, skip_special_tokens=True)[0]
print(response)# Serve with vLLM for high-throughput inference
vllm serve baidu/Qianfan-OCR --trust-remote-code --hf-overrides '{"architectures": ["InternVLChatModel"]}'Technical details
What the source and the catalog store for this checkpoint. Missing fields are left out, never zero.
- Checkpoint / source
- baidu/Qianfan-OCR
- Task
- image-text-to-text
- Architecture
- QianfanOCRForConditionalGeneration
- Library in source
- transformers
- Weight license
- Apache 2.0baidu/Qianfan-OCR model card (Hugging Face) · read 2026-10-18source
- Commercial use
- Allowed Apache License 2.0 text (SPDX) · read 2026-10-18license text
- Languages
- MultilingualModel card metadata · read 2026-10-25source
- Released
- Qianfan-VL README news: "[2026/03/18] Qianfan-OCR Released!" (GitHub, Baidu) · read 2026-10-18source
Every stored field
- Revision checked
- 623bf5d20d446abdb36606aa4547cd0c18886fe5
- Architecture
- QianfanOCRForConditionalGeneration
- Library
- transformers
- Hub task tag
- image-text-to-text
- Formats
- transformers, safetensors
- Access
- Open download
- Publisher kind
- Official
- Hub parameter count
- 4,741,408,256safetensors.total
- Repository created
- 2026-03-18creation is not the model's release
- Repository updated
- 2026-04-29a commit is not a release date
- Hub downloads (30 days) · likes
- 41,257 · 1,216this repository only · read 2026-10-25
- Catalog state
- Hub config or weights found; no inventory benchmark link
Sources
Identity and published specifications come from the checkpoint page. Benchmark scores belong to the named release and do not measure speed on your computer.
- Publisher model card
- License: baidu/Qianfan-OCR model card (Hugging Face)
- Release date: Qianfan-VL README news: "[2026/03/18] Qianfan-OCR Released!" (GitHub, Baidu)
Curated facts checked: 2026-10-18. The license and release date cite the publisher's sources; parameters, files and task come from the Hugging Face record, and memory is a labelled estimate from its parameter count. A value no source states stays empty; a repository commit is not a model release date.






























































