FireRedTeam releases FireRed-OCR-2B structured document parsing model, scoring 92.94 on OmniDocBench v1.5
Original title:FireRedTeam/FireRed-OCR
AI summary
FireRedTeam released the weights for FireRed-OCR-2B, based on Qwen3-VL-2B-Instruct, achieving an overall score of 92.94 on OmniDocBench v1.5, higher than DeepSeek-OCR 2's 91.09 and PaddleOCR-VL's 92.86.
Key Takeaways
- FireRedTeam released the weights for FireRed-OCR-2B,based on Qwen3-VL-2B-Instruct,achieving an…
- 94 on OmniDocBench v1.
- 5,higher than DeepSeek-OCR 2's 91.
- 09 and PaddleOCR-VL's 92.
Quick Facts
- Published
- Updated
- Source
- FireRedTeam:原创语音与多模态项目
- Category
- Model Releases
- AI score
- 60
原文给出了 OmniDocBench v1.5 具体分数、对比模型和三阶段训练方法,读者可以据此判断这个 2B 文档解析模型的实际位置。
Related entities
Source & copyright
This article is based on curated titles and AI summaries from the AIHOT public API, with attribution and links back. Original copyright remains with the original publishers and authors; this site does not mirror full texts.
How to Cite
When citing this article, keep the title, date, and permalink, and credit Orange News as the source.
Orange News: FireRedTeam releases FireRed-OCR-2B structured document parsing model, scoring 92.94 on OmniDocBench v1.5. February 28, 2026. https://new.nat6.net/en/news/cmtym5t890006roy8gecuetvb/