Cohere Embed v4
Cohere's multimodal embedding model: text and images in one vector space, 128K context.
About
Embed v4 is Cohere's flagship embedding model, announced April 15, 2025. It natively embeds text and images — including screenshots of PDFs, slides, tables, and figures — into a unified vector space, with a 128K context window (roughly a 200-page document), 100+ languages, and Matryoshka-truncated dimensions of 256/512/1024/1536.
Cohere positions it for enterprise RAG over messy real-world documents, with industry tuning for finance, healthcare, and manufacturing. It is a closed, API-only model, and the headline MTEB score of about 65.2 is Cohere-reported — embedding benchmark numbers are self-reported across the industry and should be treated as directional.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
- MTEB ~65.2, slightly ahead of text-embedding-3-large ~64.6 (Cohere-reported)
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 7/10
Scored 7/10 (metrics tier, metrics-v1). Evidence: Cohere flagship embedding; available on Cohere Platform, Bedrock, Azure, OCI; Closed proprietary API-only model; Cohere-reported MTEB ~65.2; embedding benchmarks are self-reported industry-wide. License: Proprietary (API terms).
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 8.0/10
- Maintenance · recency, releases, issue load
- 9.0/10
- Openness · license + docs
- 3.0/10
- Manual review · sanity read + red-flag check
- 7.0/10
- · Cohere flagship embedding; available on Cohere Platform, Bedrock, Azure, OCI
- · Closed proprietary API-only model
- · Cohere-reported MTEB ~65.2; embedding benchmarks are self-reported industry-wide