---
id: fasttext-predict
version: "0.9.2.4"
license: MIT
license_treatment: permissive
maintenance: dormant
---

# fasttext-predict — fasttext with wheels and no external dependency, but only the predict method (<1MB)

License: permissive · Maintenance: dormant · Downloads: 1.5M/mo

## What it is and what it does

fasttext-predict is a stripped-down Python wrapper around the fastText library that exposes only the model prediction interface. It is designed for language identification and similar inference tasks where you already have a trained model and only need to run predictions on text. The package removes all training, model manipulation, and other fastText features, keeping the binary under 1MB and eliminating external dependencies like NumPy.

You load a pre-trained fastText model file and call its predict method to classify or identify text. This makes it suitable for embedding language detection into applications where you want minimal dependencies and fast startup. The package provides pre-built wheels for Python 3.9–3.13 across macOS, Windows, and Linux architectures, so installation typically requires no compilation on common platforms.

Use it for:

- Detect the language of user-submitted text in a web application without pulling in a heavy ML stack.
- Identify document language in a batch processing pipeline where only inference is needed.
- Embed language detection in a lightweight CLI tool or edge service with minimal footprint.
- Classify short text snippets using a pre-trained fastText model in a real-time API.
- Add language identification to a search engine or content aggregator without external ML dependencies.

## Worth the install?

AI-flagged interpretation of the facts above — verify before relying.

Loads and runs fastText language identification models with a minimal, dependency-free predict interface.

Yes, if you need lightweight language identification on a pre-trained fastText model and want to avoid NumPy and other heavy dependencies. The dormant maintenance status (629 days since last release) is a concern for long-term support, but the package has no known vulnerabilities and the predict-only scope limits the surface for bugs. Install it for inference-only workloads; do not expect active development or new features.

## Install

pip install fasttext-predict
uv add fasttext-predict
poetry add fasttext-predict

## Installing fasttext-predict

Before you install: Medium install friction due to compiled wheels across many architectures, but no runtime dependencies. Dormant maintenance (629 days since last release) means no active development or bug fixes.

License in practice: MIT license is permissive; you may use, modify, and distribute this package freely with minimal restrictions.

Quickstart:

pip install fasttext-predict

import fasttext
model = fasttext.load_model('lid.176.ftz')
result = model.predict('Fondant au chocolat et tarte aux myrtilles')

You must provide a fastText model file (e.g., a .ftz file); the package only includes the predict method, not model training or other fastText features.

Verify before relying:

- Whether the package's predict method is compatible with all fastText model formats or only specific ones.
- Performance characteristics compared to the full fastText library for production language identification workloads.
- Whether dormant status means the package will receive security updates if vulnerabilities are discovered.

## Package facts

- License: MIT (permissive)
- Python support: unspecified
- Install friction: medium
- Maintenance: dormant
- Downloads: 1.5M/month (top 5,000 on PyPI)
- Known vulnerabilities: none known

## Tags

language identification, fasttext predict, language detection python, lightweight text classification, fasttext model inference, language-detection, inference-only, minimal-deps

[View on SkillFed](https://skillfed.io/packages/fasttext-predict) · [View on PyPI](https://pypi.org/project/fasttext-predict/)