GLiNER-2

Run Inference

Perform fast, schema-driven entity extraction, text classification, and structured data parsing through the hosted GLiNER 2 inference engine.

POST /gliner-2

Purpose

Use this endpoint to run Named Entity Recognition, Text Classification, or Structured Data Extraction without setting up local models.
GLiNER 2 unifies multiple information-extraction tasks in a single efficient forward pass optimized for CPU-based inference.

Typical use-cases include:

Extracting entities such as people, organizations, products, or places
Classifying text for sentiment, topic, or intent
Parsing structured fields from unstructured text such as invoices, contracts, or medical records

Endpoint

POST https://api.fastino.ai/gliner-2

Headers

x-api-key: pk_...
Content-Type: application/json

Request Body

Field	Type	Required	Description
text	string	Yes	The input text to analyze.
labels	object / list	Yes	Entity labels, class labels, or structured-field schema depending on task type.
threshold	float	Optional	Confidence threshold for predictions (default = 0.5).
task	string	Optional	Optional task specifier (e.g. `"extract_json"` for structured output).
schema	object	Optional	Schema definition for structured extraction (required if using `"extract_json"`).

Example 1 – Entity Extraction

{
  "text": "Apple CEO Tim Cook announced iPhone 15 in Cupertino yesterday.",
  "labels": ["company", "person", "product", "location"]
}

Example Response

{
  "entities": {
    "company": ["Apple"],
    "person": ["Tim Cook"],
    "product": ["iPhone 15"],
    "location": ["Cupertino"]
  }
}

Example 2 – Text Classification

{
  "text": "This laptop has great performance but poor battery life.",
  "labels": { "sentiment": ["positive", "neutral", "negative"] }
}

Example Response

{ "sentiment": "negative" }

Example 3 – Structured Extraction

{
  "task": "extract_json",
  "text": "iPhone 15 Pro Max with 256GB storage priced at $1199.",
  "schema": {
    "product": [
      "name::str::Product name",
      "storage::str::Storage capacity",
      "price::str::Price with currency"
    ]
  }
}

Example Response

{
  "product": [{
    "name": "iPhone 15 Pro Max",
    "storage": "256GB",
    "price": "$1199"
  }]
}

Example cURL

curl -X POST "https://api.fastino.ai/gliner-2" \
  -H "x-api-key: pk_test_123" \
  -H "Content-Type: application/json" \
  -d '{
    "text": "Apple CEO Tim Cook announced iPhone 15 in Cupertino yesterday.",
    "labels": ["company", "person", "product", "location"]
  }'

Response Fields

Field	Type	Description
entities	object	Map of detected entity labels to extracted text spans.
classification	object / string	For classification tasks, the predicted label(s).
structured_output	object	For structured extraction, key–value pairs following the provided schema.

Error Responses

HTTP Code	Error Code	Description
400	`INVALID_REQUEST`	Missing or malformed fields.
401	`UNAUTHORIZED`	Invalid or missing API key.
500	`SERVER_ERROR`	Internal inference error — retry with exponential backoff.

Example:

{
  "error": {
    "code": "INVALID_REQUEST",
    "message": "Missing required field: text"
  }
}

Best Practices

Keep text under 8 KB per call for optimal latency.
Reuse the same model session for batch requests to leverage warm caching.
Use threshold ≥ 0.7 for precision-critical domains such as finance or healthcare.
Combine GLiNER 2 output with the Fastino Personalization API to enable user-specific extraction or context enrichment.

Summary

POST /gliner-2 provides unified, production-ready information extraction through Fastino’s hosted GLiNER 2 engine — efficient, accurate, and fully compatible with LLM pipelines.

Join our Discord Community

On this page

Title