OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Models

Gliner

GLiNER is an open-source Python framework for training and deploying lightweight named-entity-recognition models. It supports zero-shot entity extraction with user-defined labels, as well as streaming NER, relation extraction, PII detection, information extraction and token classification. Models can be fine-tuned, and the framework includes a Ray Serve-based option for serving inference.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Gliner — OpenViST project profile in AI Models.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use GLiNER when a Python text pipeline needs named entities defined by the task, especially when zero-shot labels are useful or a lightweight, fine-tunable model is preferred. Its serving layer is a documented next step for higher-throughput or multi-user inference workloads.

Who it is for

  • Developers building text-processing pipelines that need to identify entities under chosen labels.
  • Teams handling text that need entity and relation extraction or PII detection.
  • Practitioners who want to adapt an NER model using labeled examples.
  • Engineers evaluating lightweight NER inference or a Ray Serve-based serving layer.

Documented uses

  • Extract categories such as people, dates, organizations or locations from supplied text.
  • Recognize entities incrementally as text arrives.
  • Extract entities and their relations together.
  • Detect personally identifiable information in text.
  • Fine-tune a model for a labeled information-extraction task.
Before you choose

Requirements & limits

The project is available as a public GitHub repository and is licensed under Apache-2.0. The documented package installation uses pip or uv.

  • The basic interface accepts labels at prediction time, which suits tasks where entity categories vary.
  • Fine-tuning is supported for tasks with labeled examples.
  • The framework is designed for CPU and consumer-hardware use; optimization options include compilation and quantization.
  • INT8 quantization requires Quantization-Aware Training to preserve accuracy, according to the README.
  • Ray Serve support is installed separately from the basic package.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Install GLiNER and extract labeled entities

Set up the Python package and use its documented basic interface to extract entities from text.

Before you start

  • Python environment with pip or uv
  • Text to process and entity labels
  • Access to load the documented pretrained model

Steps

  1. Install the package with pip install gliner, or use uv pip install gliner.

  2. Import GLiNER and load gliner-community/gliner_small-v2.5 with GLiNER.from_pretrained.

  3. Pass text and a list of labels to model.predict_entities, optionally setting a threshold.

  4. Read the returned entity text and label fields.

Expected outcome

A list of text spans identified under the requested entity labels.

Guide limitations

  • The example documents basic entity extraction; it does not establish accuracy for a particular dataset or use case.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
13 October 2026
Original website title
urchade/GLiNER · Official homepage title · retained in its source language
How we review sourcesSuggest a correction