Cohere Embed v3

Overview

Cohere Embed v3 is a text-only embedding model that generates 1,024-dimensional vectors for semantic search and retrieval applications. Available in both English-only and multilingual variants.

Key Features

• High-Quality Embeddings: Optimized for search and retrieval tasks
• Multilingual Support: Supports 100+ languages in multilingual variant
• 1,024 Dimensions: Dense vector representations
• Production-Ready: Enterprise-grade API with high availability
• Multiple Deployment Options: Available via Cohere API, AWS Marketplace, Azure Marketplace

Pricing

API Pricing

• Embed v3: $0.10 per million tokens
• Embed v4 (newer): $0.12 per million tokens (text), $0.47 per million image tokens

Free Tier

• Trial API Key: 1,000 free API calls per month
• Rate Limits: 5 calls/minute for Embed, 20 calls/minute for Chat
• Access to all models including Command R+, Rerank 3.5, and Embed 4

Billing

• Pay-as-you-go for Production keys
• Monthly billing or when reaching $250 in outstanding balances
• Priced based on tokens embedded

Model Variants

• embed-english-v3.0: English-only model
• embed-multilingual-v3.0: 100+ language support

Cloud Marketplace Availability

• AWS Marketplace: Self-service deployment
• Azure Marketplace: Managed service options

Use Cases

• Semantic search applications
• Document retrieval systems
• RAG pipelines
• Content recommendation
• Clustering and classification

API Access

Available through:

• Cohere REST API
• Python SDK
• Node.js SDK
• Cloud marketplace deployments