Sub-1B open OCR models benchmarked: Surya 2.5 tops OmniParseBench at 85.9
VikParuchuri · x · 2026-11-04
Datalab benchmarked the recent wave of small open-source OCR models (all under 1B parameters) on OmniParseBench, using vendor-recommended settings and a headline score averaged over text, tables and layout (scores dated 2026-11-03):
- Surya 2.5 (Datalab, 0.69B): 85.9, first place
- OvisOCR3 (ATH-MaaS, 0.85B): 84.2
- LightOnOCR-3-0.8B (LightOn, 0.85B): 81.6
- Nemotron 3.5 Parse (NVIDIA, 0.63B): 80.0
- PaddleOCR-VL-1.6 (PaddlePaddle, 0.96B): 78.7
- Xiaomi-OCR-0 (0.87B): not scored — no layout output, so no headline score
Related event: OmniParseBench Launches: Surya 2.5 Tops Small OCR Model Ranking(2 posts)→
More from Models
- Gemini 4 Carbon takes #1 on Artificial Analysis Intelligence Index — thursdai_pod · 2026-11-04
- Open decision model REAX 1.3 launches: €0.40 per million input tokens, output free — airesearch12 · 2026-11-04
- REAX 1.3 ships open weights under MIT license with two miner-trained models inside — airesearch12 · 2026-11-04
- Anthropic faces questions over Claude's rules on AI-generated video game clones — emmanuelvivier · 2026-11-04
- Simon Willison replicates GPT-6 Astra weight-interpretation feat with Qwen 3.8 27B in 45 minutes — cephaloform · 2026-11-04
- Qwen, GPT and Gemini Flash all fail the same way on this service — matdryhurst · 2026-11-04