Embed 5 Overview

Name: Embed 5
Brand: Embed 5

Embed 5 is a Large Language Models (LLMs) tool. State-of-the-art embedding models for enterprise search, RAG, and document retrieval. Key features include Dual-Tier Architecture, Multimodal Document Processing, and Extended Context Window. Best for data scientists and analysts, software developers and engineers and financial advisors and analysts.

6 key features6+ alternatives →

About Embed 5

Embed 5 is Cohere's most powerful embedding family, delivering frontier retrieval quality across multimodal, multilingual, and financial documents. Available in Pro and Fast tiers that share one embedding space for flexible enterprise workflows.

Key Features

Dual-Tier Architecture.

Embed 5 Pro optimizes for maximum retrieval quality while Embed 5 Fast delivers low-latency performance. Both models share a single embedding space, so you can index with Pro and query with Fast without rebuilding your index.

Multimodal Document Processing.

Handles text, images, and mixed text-image inputs like PDF pages, scanned documents, charts, and tables. Turns visually rich enterprise content into searchable representations of meaning, even when exact words don't match.

Extended Context Window.

Supports up to 128,000 tokens, allowing you to process long documents without awkward chunk splits. Particularly useful for contracts, annual reports, service documentation, and technical manuals.

Multilingual Capabilities.

Works across more than 100 languages with support for cross-language search. Enables global teams to build retrieval systems that work seamlessly across different language documents and queries.

Financial Document Excellence.

Achieves industry-leading performance on financial content including company filings, annual reports, earnings materials, and spreadsheets. Scores highest on FinanceBench and FinQA benchmarks among tested models.

Flexible Output Formats.

Offers Matryoshka embeddings with dimensions ranging from 256 to 2,048, plus float, int8, and binary output formats. Gives you control over storage costs and retrieval speed based on your specific needs.

Frequently Asked Questions

Embed 5 is a family of embedding models from Cohere that converts text, images, and documents into searchable vector representations. It powers enterprise search, RAG systems, and AI agents by finding relevant information before it reaches generative models. The Pro tier focuses on quality while the Fast tier optimizes for speed and cost.

Embed 5 Pro costs $0.12 per million text tokens and Embed 5 Fast costs $0.08 per million text tokens. Image inputs cost $0.40 per million tokens for either model. You can index documents with the more expensive Pro model and handle queries with the cheaper Fast model to optimize costs.

Yes. Both models share the same embedding space, which means you can build your index once with Pro and then query it with Fast without regenerating vectors. Cohere's testing shows this combination achieves 98.4% of Pro-only performance while reducing query costs by 33%.

Embed 5 is available through the Cohere API, Model Vault, Microsoft Foundry, Amazon SageMaker, and directly in North. You can deploy it via shared API access or use single-tenant deployment through Model Vault for private VPC or on-premises installations.