01 What happened
Cohere released Rerank 5 with Context Focus and calibrated relevance scores, in Pro and Fast tiers with pricing from $0.02 per 1M tokens.
02 Key details
- Cohere released Rerank 5 Pro and Rerank 5 Fast, with Pro leading public benchmarks such as ViDoRe V3, nanoMTEB, BRIGHT and BEIR.
- Context Focus is a new optional capability that returns scored text spans within documents and can be configured with thresholds to reduce LLM tokens.
- Scores are calibrated as probabilities with about 0.06 expected calibration error; API pricing is $0.05 per 1M tokens for Pro and $0.02 for Fast.
- General availability includes the Cohere API, Model Vault, Microsoft Foundry, Amazon SageMaker and North; private VPC/on-premises deployment needs Cohere contact.
03 Why it matters
Rerank 5 aims to lower retrieval token cost while preserving answer quality, with two speed tiers and broad platform availability.
04 Who it matters to
Software developers, AI agent developers, RAG system developers, AI integration specialists and organizations with frequent repeatable requests.
Original source