Skip to content

Additional Embedders

Prompt Amplifier v0.2.0 adds support for several new embedding providers beyond the core TF-IDF, BM25, Sentence Transformers, OpenAI, and Google embedders.

Cohere Embeddings

High-quality embeddings from Cohere with built-in reranking support.

from prompt_amplifier.embedders import CohereEmbedder

# Basic usage
embedder = CohereEmbedder(
    api_key="your-cohere-api-key",  # or COHERE_API_KEY env var
    model="embed-english-v3.0",
)

result = embedder.embed(["Your text here"])
print(f"Dimension: {result.dimension}")  # 1024

Cohere Reranker

Improve retrieval quality by reranking results:

from prompt_amplifier.embedders import CohereRerankEmbedder

reranker = CohereRerankEmbedder(
    api_key="your-cohere-api-key",
    model="rerank-english-v2.0",
)

# Rerank retrieved documents
query = "What is machine learning?"
documents = [
    "ML is a subset of AI",
    "Weather forecast for tomorrow",
    "Deep learning uses neural networks",
]

reranked = reranker.rerank(query, documents, top_n=2)
for doc, score in reranked:
    print(f"{score:.3f}: {doc}")

Installation

pip install prompt-amplifier[embeddings-cohere]

Voyage AI Embeddings

Specialized embeddings optimized for retrieval and RAG applications.

from prompt_amplifier.embedders import VoyageEmbedder

embedder = VoyageEmbedder(
    api_key="your-voyage-api-key",  # or VOYAGE_API_KEY env var
    model="voyage-2",  # or voyage-lite-02-instruct
)

result = embedder.embed([
    "Document for indexing",
    "Another document",
])

Available Models

Model Dimension Best For
voyage-2 1024 General purpose
voyage-large-2 1536 Maximum quality
voyage-code-2 1536 Code retrieval
voyage-lite-02-instruct 1024 Fast inference

Installation

pip install prompt-amplifier[embeddings-voyage]

Jina AI Embeddings

Multilingual embeddings with excellent performance across 100+ languages.

from prompt_amplifier.embedders import JinaEmbedder

embedder = JinaEmbedder(
    api_key="your-jina-api-key",  # or JINA_API_KEY env var
    model="jina-embeddings-v2-base-en",
)

# Multilingual example
result = embedder.embed([
    "Hello world",
    "Hallo Welt",
    "Bonjour le monde",
])

Available Models

Model Languages Dimension
jina-embeddings-v2-base-en English 768
jina-embeddings-v2-base-de German 768
jina-embeddings-v2-base-multilingual 100+ 768
jina-embeddings-v2-small-en English 512

Installation

pip install prompt-amplifier[embeddings-jina]

Mistral AI Embeddings

European AI embeddings with strong multilingual capabilities.

from prompt_amplifier.embedders import MistralEmbedder

embedder = MistralEmbedder(
    api_key="your-mistral-api-key",  # or MISTRAL_API_KEY env var
    model="mistral-embed",
)

result = embedder.embed(["Your text here"])
print(f"Dimension: {result.dimension}")  # 1024

Features

  • GDPR compliant (EU-based)
  • Strong multilingual support
  • Competitive pricing
  • Fast inference

Installation

pip install prompt-amplifier[embeddings-mistral]

FastEmbed (Local)

Fast, local embeddings using ONNX runtime - no API keys needed!

from prompt_amplifier.embedders import FastEmbedEmbedder

embedder = FastEmbedEmbedder(
    model_name="BAAI/bge-small-en-v1.5",
)

# Runs completely locally
result = embedder.embed([
    "This runs on your machine",
    "No API calls needed",
])

Available Models

Model Dimension Speed
BAAI/bge-small-en-v1.5 384 Fastest
BAAI/bge-base-en-v1.5 768 Balanced
BAAI/bge-large-en-v1.5 1024 Best quality

Installation

pip install fastembed

Comparison Table

Provider Local Cost Quality Speed Multilingual
TF-IDF ✅ Free ⭐⭐ ⭐⭐⭐⭐⭐ ❌
BM25 ✅ Free ⭐⭐⭐ ⭐⭐⭐⭐⭐ ❌
Sentence Transformers ✅ Free ⭐⭐⭐⭐ ⭐⭐⭐ ✅
FastEmbed ✅ Free ⭐⭐⭐⭐ ⭐⭐⭐⭐ ✅
OpenAI ❌ $$ ⭐⭐⭐⭐⭐ ⭐⭐⭐⭐ ✅
Google ❌ $$ ⭐⭐⭐⭐⭐ ⭐⭐⭐⭐ ✅
Cohere ❌ $$ ⭐⭐⭐⭐⭐ ⭐⭐⭐⭐ ✅
Voyage ❌ $$ ⭐⭐⭐⭐⭐ ⭐⭐⭐⭐ ✅
Jina ❌ $$ ⭐⭐⭐⭐ ⭐⭐⭐⭐ ✅✅
Mistral ❌ $$ ⭐⭐⭐⭐ ⭐⭐⭐⭐ ✅✅

Choosing an Embedder

For Research/Prototyping

# Free, local, good quality
from prompt_amplifier.embedders import SentenceTransformerEmbedder
embedder = SentenceTransformerEmbedder()

For Production (Budget)

# Fast, local, ONNX optimized
from prompt_amplifier.embedders import FastEmbedEmbedder
embedder = FastEmbedEmbedder()

For Production (Quality)

# Best quality, paid API
from prompt_amplifier.embedders import CohereEmbedder
embedder = CohereEmbedder()

For Multilingual

# Best multilingual support
from prompt_amplifier.embedders import JinaEmbedder
embedder = JinaEmbedder(model="jina-embeddings-v2-base-multilingual")

For GDPR Compliance

# EU-based provider
from prompt_amplifier.embedders import MistralEmbedder
embedder = MistralEmbedder()

# Optimized for code
from prompt_amplifier.embedders import VoyageEmbedder
embedder = VoyageEmbedder(model="voyage-code-2")