OCR models / INF Technology

Infinity-Parser2 Pro

Name: Infinity-Parser2 Pro
Author: INF Technology

INF Technology · released 2026-05-11 · data checked 2026-11-06

Infinity-Parser2 Pro is a 35.1B-parameter OCR and document-parsing model from INF Technology (CN), released 2026-05-11 under the Apache-2.0 license. It scores 93.95 on OmniDocBench v1.6 (#14 of 33) and 87.6 on olmOCR-Bench (#2 of 24). It needs about 71.2 GB of memory at 16-bit.

Mixture-of-Experts on Qwen3.5 (35B total, about 3B active). Mainly English and Chinese.

OmniDocBench v1.6

93.95

#14 of 33 models

olmOCR-Bench

87.6

#2 of 24 models

Parameters

35.1B

3B active · Mixture-of-Experts

Memory to run

~71.2 GB

16-bit weights plus about 1 GB

Benchmarks

OmniDocBench v1.6 Lab-reported93.95

Best: OXR 1.0 97.03

olmOCR-Bench Lab-reported87.6

Best: OXR 1.0 87.8

The tick marks the best score on each benchmark. See how to read the ranking.

Specifications

Lab: INF Technology (CN)
Type: OCR / document parsing specialist
Total parameters: 35.1B
Active parameters: 3B
License: Apache-2.0
Memory to run: ~71.2 GB at 16-bit
Released: 2026-05-11

Weights and usage on Hugging Face ↗

Closest alternatives

Unlimited-OCR (3.3B, 94.00 on OmniDocBench, MIT)
Qianfan-OCR (4.7B, 93.90 on OmniDocBench, Apache-2.0)
Youtu-Parsing (2.5B, 93.74 on OmniDocBench, Youtu-Parsing license)
Ovis2.6 30B-A3B (31.4B, 93.70 on OmniDocBench, Apache-2.0)

Sources

https://huggingface.co/infly/Infinity-Parser2-Pro

Something wrong here? Report it.