# Is MiniCPM-OCR 1.0 SOTA?

Pending benchmarks. Announced Oct 27, 2026; no independent results yet.

As of Nov 3, 2026, MiniCPM-OCR 1.0 is pending benchmarks on isitsota: announced Oct 27, 2026, with no independent results on the boards we track yet.

OpenBMB's document-parsing model, the first OCR-specific model in the MiniCPM-V/o family. It reads text, tables, formulas, charts and seals in Chinese and English documents and PDFs, including scans, photos and curved or skewed pages, and extracts key information. Open weights under Apache 2.0.

Canonical: https://isitsota.com/m/minicpm-ocr-1-0
Last updated: 2026-10-27
Tracking since: 2026-10-27

## Announced facts

| Fact | Value |
| --- | --- |
| Lab | OpenBMB |
| Announced | Oct 27, 2026 |
| API | Not yet · Online demo on ModelBest's platform; the lab links a document-parse API, but its GitHub API guide says the official hosted endpoint is still to be added (unsure) |
| Open weights | Published · Hugging Face openbmb/MiniCPM-OCR-1.0, plus GGUF, AWQ, GPTQ and DSpark (speculative decoding) builds |
| Parameters | 1.3B |
| Base | Trained for OCR on top of MiniCPM-V 4.6 (SigLIP vision encoder + Qwen3.5 language model) |
| Languages | Chinese and English |
| Pipeline | PP-DocLayoutV3 layout model + MiniCPM-OCR 1.0 + minicpmvocr SDK; outputs Markdown or JSON |
| Runtimes | vLLM, SGLang, llama.cpp and Ollama |
| License | Apache 2.0 |

Tags marked unsure are not stated on the lab's own pages.

## What would let us call it

Verdicts use only independent boards we already track. Two consecutive snapshots are required before any crown.

- [x] Publicly announced with a source
- [x] Usable by the public (API or weights)
- [ ] Listed on an independent board we track
- [ ] Enough results for a rank in a category
- [ ] Holds across two snapshots (crowns only)

## What the lab claims

Self-reported. Shown for context; not used in any verdict.

- Lab claim: 93.68 on Real5-OmniDocBench and 54.7 on OCRBench v2, with the highest equal-weight average across OmniDocBench v1.6, Real5-OmniDocBench, Wild-OmniDocBench and olmOCR-Bench among the open OCR models OpenBMB compared.
- Lab claim: 1.2 pages/s on an RTX 5090; with DSpark about 1.49x the throughput of PaddleOCR-VL-1.6 and 3.62x MinerU2.5 Pro.

## Sources

- LAB: [OpenBMB announcement on X](https://x.com/OpenBMB/status/2115124474882469955) (2026-10-27)
- REPO: [OpenBMB/MiniCPM-OCR on GitHub](https://github.com/OpenBMB/MiniCPM-OCR) (2026-10-27)
- REPO: [openbmb/MiniCPM-OCR-1.0 on Hugging Face](https://huggingface.co/openbmb/MiniCPM-OCR-1.0)

Tracking continues until an independent board lists it.

## Related

- Lab: [OpenBMB](https://isitsota.com/labs/openbmb.md)
- All pending releases: https://isitsota.com/releases.md
- JSON: https://api.isitsota.com/v1/models/minicpm-ocr-1-0