starter.wiki

Managed retrieval model APIs

Cohere Embed and Rerank

Managed embedding and reranking APIs for text, multilingual, image, and structured retrieval workloads under Cohere's current model lifecycle.

Visit official site

Editorial verdict

Choose Cohere when corpus-specific evaluation shows that its embedding and optional reranking route produces useful relevance within supported inputs, languages, dimensions, throughput, data, and cost boundaries.123

Best for

  • Teams evaluating embedding and reranking together
  • Multilingual or multimodal retrieval workloads supported by current models
  • Products prepared to monitor model deprecations and re-embedding cost
123

Not ideal for

  • Teams that only need the provider already used for generation
  • Workloads without a representative retrieval evaluation set
  • Systems unable to re-embed or migrate when models retire
123
Main trade-off

Retrieval-focused models and reranking can improve relevance while adding a model provider, extra request stage, rate limits, token cost, evaluation work, and migration risk.123

Product boundary

Whether Cohere's current Embed and Rerank models improve retrieval enough on the actual corpus to justify their model, rate-limit, price, and migration boundaries.

This page owns Cohere Embed and Rerank selection, not the full Command generation platform, vector storage, chunking architecture, or generic RAG implementation.123

For: Search and RAG teams evaluating embeddings and optional reranking on representative documents and queries

  • Another embedding family performs better on the real corpus
  • Reranking does not justify its latency and cost
  • Model lifecycle, rate limits, data, or region boundaries do not fit

Why teams consider Cohere Embed and Rerank

  • Embed routesCohere documents text and image embedding models with task types, dimensions, and supported inputs.123
  • Rerank stageRerank models can reorder retrieved candidates without replacing the underlying index.123
  • Lifecycle visibilityOfficial documentation publishes current rates, production-key limits, and model deprecations.123

Pricing

Embedding is billed by processed tokens and reranking by its current processed-search or token metric. Trial and production keys have separate usage and rate-limit boundaries.3

Current decision boundary

Usage-based Embed and Rerank APIs

Verified 2026-07-27: Cohere publishes model-specific Embed and Rerank pricing; estimates must include index creation, query embedding, reranking volume, batch behavior, and re-embedding.3

Embedding meter
Processed input tokens by selected model3
Rerank meter
Current model-specific processed request or token dimensions3
Migration cost
Corpus re-embedding and retrieval re-evaluation3
Pricing checked View official pricing

Cohere Embed and Rerank vs alternatives

OpenAI API

Choose when
Products needing a broad managed model and tool surface
Avoid when
Teams requiring self-operated model weights or another cloud control plane
Compared with Cohere Embed and Rerank
A broad managed platform reduces model-serving and integration work while increasing dependence on OpenAI-specific models, APIs, pricing meters, policies, and migration schedules.567

Voyage AI

Choose when
Teams optimizing retrieval relevance on domain-specific corpora
Avoid when
Teams prioritizing one broad generation provider over specialized retrieval
Compared with Cohere Embed and Rerank
Specialized retrieval models may improve relevance or efficiency while adding another provider, model lifecycle, re-embedding, reranking, data, and distribution dependency.8910

Google Gemini API

Choose when
Applications benefiting from Gemini's current multimodal model surface
Avoid when
Teams that need Vertex-specific governance but are evaluating only the Developer API
Compared with Cohere Embed and Rerank
A broad Google model surface and one SDK reduce integration work while free-versus-paid data terms, model stages, quotas, pricing, and Vertex separation require active governance.111213

Resources and sources

Official product, pricing, policy, and lifecycle sources

  • Cohere Embed models
    Open
  • Cohere Rerank models
    Open
  • Cohere pricing
    Open
  • Cohere model deprecations
    Open

Related tasks

  1. 1
    Cohere Embed models

    Cohere · Accessed Official

  2. 2
    Cohere Rerank models

    Cohere · Accessed Official

  3. 3
    Cohere pricing

    Cohere · Accessed Official

  4. 4
    Cohere model deprecations

    Cohere · Accessed Official

  5. 5
    OpenAI API model catalog

    OpenAI · Accessed Official

  6. 6
    OpenAI API pricing

    OpenAI · Accessed Official

  7. 7
    OpenAI API data controls

    OpenAI · Accessed Official

  8. 8
    Voyage AI text embeddings

    Voyage AI · Accessed Official

  9. 9
    Voyage AI pricing

    Voyage AI · Accessed Official

  10. 10
    Voyage AI joining MongoDB

    Voyage AI · Accessed Official

  11. 11
    Gemini API models

    Google · Accessed Official

  12. 12
    Gemini Developer API pricing

    Google · Accessed Official

  13. 13
    Gemini API additional terms

    Google · Accessed Official