Vision, speech & language

AI Ability APIs

260+ production-grade APIs that solve a specific job — extract, read, hear, speak, detect — ready to integrate in minutes.

PP-OCRv6

Latest enterprise OCR across 50 languages. Polygon coordinates, confidence scores, orientation and unwarping. Public beta includes 2,000 free calls for enterprises.

PaddleOCR

PP-StructureV3 document parsing for layouts, tables, formulas and reading order, plus PaddleOCR-VL. Open source and privately deployable.

Managed OCR

General text — printed and handwritten — plus forms, tables, receipts, invoices, IDs, passports and bank cards.

Speech, NLP & vision

ASR and TTS, five-second voice cloning, sentiment, classification, NER, object detection and face recognition.

Offline SDKs are available for identity OCR, so documents can be processed on-premise with no cloud dependency. PaddleOCR is open source at github.com/PaddlePaddle/PaddleOCR.
Start

Not sure which line fits?

Tell us the workload — text, video, location or a full product — and we will scope the stack, the region and a firm rate.