Facebook AI Research

fastText

Efficient word representations with subword information.

Handles out-of-vocabulary wordsFast training and inferenceGood for low-resource languagesSupports text classification
Today's score
87.0
Try fastText

Where it ranks today

Best for / Not great for

Best for
  • Text classification
  • Handling typos and rare words
  • Low-resource language NLP
  • Baseline word embeddings
Not great for
  • Contextual word embeddings
  • Advanced semantic search
  • Complex sentence/document understanding
  • Modern RAG pipelines

Why it ranks here

fastText's ability to handle subword information and out-of-vocabulary words keeps it relevant for specific tasks like text classification and low-resource languages. However, its lack of contextual understanding limits its utility in modern semantic search and RAG.

30-day trend

Score breakdown

Search trends86
Benchmarks87
Developer buzz88
News mentions85

Pricing

API: $0.00 in · $0.00 out per 1M tokens · Consumer: $0.00/mo

Pricing plans

Popular
Pre-trained Models (Open Source)
Downloadable models for various languages.
Free
  • Subword information
  • Fast training
  • Support for many languages
  • Text classification integration
Download models
Compare with another modelHow is this score calculated? →Snapshot 2026-08-13