Dharma-AI Releases Specialized 2B Parameter Field Extraction Model
In short: Dharma-AI has introduced a 2-billion-parameter model specialized in extracting structured JSON fields from various documents. It supports both open extraction and schema-directed modes, consistently outperforming frontier LLMs on field-level F1 benchmarks while significantly reducing operational costs.
This summary was generated automatically by AI from Hugging Face's publication. It is our own text, not a copy of the original — facts, figures and quotes belong to the source, linked above and below.
What changed?
- 1Released specialized 2B parameter extraction model based on DharmaOCR-LITE v1.1.
- 2Supports schema-directed extraction for specific JSON outputs.
- 3Handles open extraction where the model identifies fields contextually.
- 4Achieved 0.834 F1 score on benchmarks, exceeding frontier model performance.
| Parameter | Before | Now |
|---|---|---|
| Parameter count | Not specified | 2 billion |
Why it matters
For companies processing high volumes of invoices, IDs, or forms, this model offers a cost-effective, self-hosted alternative to expensive general-purpose APIs. It ensures structured, typed output that integrates directly into downstream systems without custom mapping code.
What it means for AI agents and contact centers
If your workflows involve high-volume document processing, this model is worth testing to compare against current API-based extraction. Focus your tests on latency, schema adherence, and performance with multi-page or complex documents. Since it is self-hosted, evaluate your infrastructure requirements (L40S or similar) against your current API spend.
🧪 Worth Testing
It offers higher quality and lower operational costs compared to general-purpose frontier models for document processing tasks.
Sources
- Hugging FaceOfficialPrimary sourceOriginal article →„Dharma Field Extractor: Any Document In, Structured Fields in JSON Out“4 Nov 2026, 19:50
- Published by source
- 4 Nov 2026, 19:50
- Found by our system
- 4 Nov 2026, 20:06
- Summary generated
- 4 Nov 2026, 20:07
This article was written by AI from the original source. Facts, numbers and prices come from the source; missing values are marked “Not specified”. Legal notice, copyright and privacy