Cohere
Builds the Command (R, R+, A) and Embed model families for enterprise NLP. Optimized for RAG, semantic search, reranking, and multi-step tool use with context windows up to 256K tokens. Supports fine-tuning on proprietary data and deployment across AWS, Azure, GCP, or on-premises VPCs for data-residency requirements.