Skip to content

AI model by Cohere

Cohere Embed v4 for business

Cohere's multimodal embedding model for enterprise search, embedding text, images and mixed documents, with flexible vector sizes and long inputs.

  • Reviewed on September 24, 2026
  • Cohere

Capability tiersRelative, not benchmarks

Reasoning Very low
Coding Very low
Vision High
Speed High
Price level Medium
Context size High
Tool calling Very low
Structured output Very low

Key facts about Cohere Embed v4

Model ID at review
embed-v4.0
Provider
Cohere
Main category
Embeddings
Open weights
No, available as a hosted service
Last reviewed
September 24, 2026

Fit

Where it fits and where it does not

Good at

  • Enterprise search across text and images.

  • Embedding mixed documents such as slides and PDFs with images.

  • Long inputs, up to 128K tokens.

  • Flexible vector sizes from 256 to 1,536.

  • Pairing with Cohere Rerank.

Not the right choice for

  • Generating answers on its own.

  • Self-hosting, since it is not open weight.

  • Projects that cannot re-embed if they switch later.

  • Audio or video.

Use cases

Business use cases we would use it for

Document search

Search by meaning across policies, contracts and reports. See knowledge base search.

Slide and PDF search

Finding content inside slides and PDFs with images.

Assistant retrieval

The search step behind a knowledge assistant.

Our notes

When we would choose it

Embed v4 is Cohere's current embedding model. It accepts text, images and mixed text-and-image inputs, handles inputs of up to 128K tokens and produces vectors from 256 to 1,536 dimensions. Cohere positions it for multimodal enterprise search and retrieval.

The long input limit and mixed inputs are its standout features. Many business documents are not plain text: slides, scanned forms and PDFs with charts. Embedding them directly, rather than extracting text first, can improve search on exactly the documents that are hardest to handle.

In a knowledge assistant, we would pair Embed v4 with Cohere Rerank and a language model such as Command A+ or Claude Sonnet. Keyword search alongside embeddings helps with product codes and names.

For comparison, Gemini Embedding 2 also embeds images and adds audio and video, while Voyage 4 and OpenAI embeddings focus on text. Test two on your own questions before choosing.

We did not verify pricing on Cohere's own site at review time, so the price level is an estimate. Confirm it before estimating the cost of embedding a large collection. A small pilot on a few thousand real documents gives a reliable cost and quality estimate before you commit to the full archive. See embedding models.

Before you commit

Things to check before you commit

  • Retrieval test

    Measure on real questions against your current search.

  • Vector size

    Choose a dimension that balances quality and storage.

  • Pricing

    We saw pricing only on third-party sites. Check Cohere directly.

  • Switching cost

    Changing models later means re-embedding everything.

Alternatives

Models to compare it with

Cohere Rerank 4

Cohere's multilingual rerank models, which sort search results by relevance, in a best-quality Pro version and a low-latency Fast version.

Gemini Embedding 2

Google's current embedding model, multimodal: it embeds text, images, video, audio and PDFs for search and retrieval.

Voyage 4

Voyage AI's current embedding series, now part of MongoDB, with large, standard and lite models that share one embedding space.

Keep exploring

FAQ

Questions people ask us

Have a question that is not here? Ask us directly.

Start a project

Not sure which model fits? Ask us to evaluate your use case.

Send a short brief. We reply with questions, a suggested plan and an estimate you can compare with other offers.

Your privacy choices

We use necessary storage to run this site. With your permission we also use Google Analytics to see which pages help people, and load maps from Google. You can change this at any time. Read the cookie policy.