OCR and document
PaddleOCR-VL-1.6
Baidu compact document-parsing VLM; top of OmniDocBench v1.6 in its own paper.
Levelmodel
ClassOCR and document
Parameters900M
Contextnot in the notes
Released2 Jun 2026
Opennessunknown
Runs onsingle datacentre GPU, CPU server, embedded GPU
Statusactive
Benchmark Results
vendor-reported
OmniDocBench v1.6multimodal
overallvendor-reported
96.33
050100
Score in score. Higher is better.
PaddleOCR-VL-1.6 paper table
Real5-OmniDocBenchmultimodal
overallvendor-reported
93.19
050100
Score in score. Higher is better.
PaddleOCR-VL-1.6 paper table
Where this comes from
Every score names who ran it: vendor-reported, maintainer-run or independent. A rank is shown with the population it was ranked in (gap rule 3).
Lineage
finetunesuccessor
ERNIE-4.5-0.3B is the parent (finetune)high confidencemodel cardPaddleOCR-VL-1.5 is the parent (successor)high confidencemodel cardWhere this comes from
Parent and child relations from model cards, Hugging Face base_model fields and the genealogy tables of the research notes. The tree shows two generations each way, 4 models. Confidence and evidence are the notes' own.
Identity
Where this comes from
Developer and dates from the research notes. Release date is the vendor's public announcement; weights date is when open weights were published.
Architecture and Size
- Architecture
- autoregressive Compact VLM: NaViT-style native-resolution encoder + adaptive MLP connector + ERNIE-4.5-0.3B decoder; PP-DocLayoutV3 layout stage
- Total parameters
- 900M sources
- Active parameters
- –
- Checkpoint parameters
- –
- Context length
- not in the notes
- Input
- image
- Output
- text
Where this comes from
Parameter counts as the developer states them. Where the checkpoint holds a different total, both are shown (register A16).
Openness and Access
- Openness
- unknown
- Licence
- Apache License 2.0 sources
- Access routes
- open weights download
Where this comes from
The licence file governs over registry tags and aggregator labels.
Deployment
- Runs on
- single datacentre GPU, CPU server, embedded GPU
- Footprint
- Single GPU; small enough for CPU/edge with effort (judgement)
- Runtimes
- not in the notes
Where this comes from
Deployment tiers and runtimes as named in the notes. Footprint is the notes' figure, not a measurement by AURION.
Indonesia
- Indonesian support
- not stated
- Regional languages
- not in the notes
- Jakarta hosting
- unknown
- Languages
- not in the notes
Where this comes from
Indonesia fields are carried on every model; unknown is an allowed value.
Industry Use Cases
No industry use case links this model.
Links
DeveloperBaidubig tech, China
Baidubig tech, ChinaModelPaddleOCR-VL-1.6Baidu, OCR and document
PaddleOCR-VL-1.6Baidu, OCR and documentR&D TopicsDocument Parsing Vision-Language Models
Use CaseseKYC Document ExtractionFinanceDocument and Archive ProcessingPublic SectorUse cases of models in the same R&D topic; none is recorded for this model itself.
Who Runs It
PT Bank Tabungan Negara (BTN)In production, 1 Jul 2026
GrabIn production, 4 Nov 2025
Komisi Pemilihan Umum (KPU)In production, 4 Mar 2024Deployments of these use cases; none names this model.
PT Bank Tabungan Negara (BTN)In production, 1 Jul 2026GrabIn production, 4 Nov 2025Komisi Pemilihan Umum (KPU)In production, 4 Mar 2024Deployments of these use cases; none names this model.Organisations 1
Sources
- arxiv.org/pdf/2606.03264 https://arxiv.org/pdf/2606.03264
- arxiv.org/pdf/2510.14528 https://arxiv.org/pdf/2510.14528
- huggingface.co/PaddlePaddle/PaddleOCR-VL-1.6 https://huggingface.co/PaddlePaddle/PaddleOCR-VL-1.6
To cite this page: Project AURION, AI Use-Cases Research & Innovation, "PaddleOCR-VL-1.6", https://aurion.semai.id/models/baidu-paddleocr-vl-1-6/, records of October 2026.