Rerank 5 is a new retrieval optimization model featuring sharper search and fewer tokens. It also introduces a new feature called Context Focus. This is Cohere's latest Rerank model release.

Cohere released Rerank 5, their most advanced reranking models for enterprise retrieval, in two variants: rerank-v5.0-pro for highest ranking quality on complex queries and rerank-v5.0-fast for low latency workloads. Key new features include Context Focus (returning relevant character spans within documents), calibrated relevance scores, multilingual support for 100+ languages, a 32k token context window, and semi-structured data support.

Cohere introduced North 2, described as 'AI without compromises' and an enterprise AI platform update. The page also highlights new models including Embed 5 (a new family of frontier embedding models), North Small Translate (a new machine translation model), Transcribe (a speech recognition model), and Parse (a document parsing model).

Cohere released Embed 5, described as a new family of frontier embedding models with state-of-the-art retrieval capabilities. It is now available in Pro and Fast tiers, representing a significant update to Cohere's semantic representation model lineup.

Cohere released Embed v5 on September 30, 2026, as indicated by the changelog entry and URL slug 'embed-v5'. This represents a new version of Cohere's embedding model with updated key features and availability details.

Cohere released Embed 5, their most powerful embeddings family yet, with major improvements over Embed 4 on visually rich documents, financial filings, parsed PDFs, code, and multilingual retrieval. The release includes two variants: embed-v5.0-pro for highest retrieval quality and embed-v5.0-fast for low latency and high throughput, sharing a common embedding space. New features include multimodal inputs, 128k token context window, support for 100+ languages, and Matryoshka embeddings with flexible dimensions.

Cohere released North Small Translate, an open-weights mixture-of-experts model purpose-built for machine translation across more than 50 languages and locale variants. The model features 218 billion total parameters with 25 billion active parameters and a 16K context length. It is available through the free-tier Chat V2 API, as FP8 open weights for non-commercial use, and through Model Vault with a commercial license.

Cohere released North Small Translate, an open-weights mixture-of-experts model purpose-built for machine translation across more than 50 languages. It is designed for researchers, developers, and enterprises to flexibly evaluate and deploy machine translation while retaining control over their data.

Cohere released Parse (parse-v5.0), a 2.3B-parameter multimodal model that converts complex documents into clean, structured Markdown for downstream AI workflows. It extracts text in reading order, tables, lists, forms, images and captions, page boundaries, and visual element locations. Outputs include Markdown/HTML content, HTML-formatted tables, bounding boxes, and image descriptions.

Cohere released Cohere Transcribe Arabic, an open-source speech-to-text model fine-tuned from Cohere Transcribe using Arabic speech data. It supports all major Arabic dialects and English spoken with an Arabic accent, and is described as the most accurate open-source Arabic ASR model available. It is optimized for production inference and throughput with a 2B parameter conformer-based encoder-decoder architecture.

Cohere released North Mini Code, their first agentic coding model. It is a 30 billion total / 3 billion active parameter Mixture of Experts model trained specifically for agentic coding tasks. The model is small enough to run on local hardware and is available via the Chat V2 API and as open weights on Hugging Face under Apache 2.0 license.