New Function

Instructor Embeddings

Instruction-tuned embeddings for precise task control.

Fine-grained control via instructionsHigh performance on specific tasksAdaptable to diverse classification needOpen-source library
Today's score
87.0
Try Instructor Embeddings

Where it ranks today

Best for / Not great for

Best for
  • Task-specific embeddings (e.g., classification, retrieval)
  • Zero-shot or few-shot learning scenarios
  • Controlling embedding dimensions precisely
  • Building custom classification systems
Not great for
  • General-purpose semantic search without instruction tuning
  • Users unfamiliar with instruction prompting
  • Extremely high-volume, low-cost scenarios requiring standard models

Why it ranks here

Instructor embeddings offer a unique approach by allowing users to guide embedding generation with natural language instructions. This fine-tuning capability makes them particularly strong for classification tasks and scenarios requiring highly specific embedding outputs, differentiating them from more general-purpose models.

30-day trend

Score breakdown

Search trends85
Benchmarks86
Developer buzz92
News mentions85

Pricing

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

Pricing plans

Popular
Free (Self-hosted)
Open-source library for instruction-based embeddings
Free
  • Python library
  • Instruction-based fine-tuning
  • Supports various underlying models
  • Full control
Install library
Compare with another modelHow is this score calculated? →Snapshot 2026-08-03