Command A
An open weights LLM you can self-host with Transformers, with document citations, tool calling and a CC-BY-NC-4.0 noncommercial license.
Command A is an open weights language model from Cohere and Cohere Labs for researchers and developers building self-hosted chatbots, document assistants and AI agents. Its focus is business tasks that combine multilingual text, supplied documents and external tools. You can run the model on your own hardware; Cohere also offers hosted chat through a playground and Hugging Face Space.
The model can answer questions using document snippets and cite the passages behind individual claims. It can also call APIs, query databases or use search engines through tools supplied by an application. Citations can point to tool results as well as documents, so readers can check where an answer's details came from.
Beyond conversational replies, Command A handles summarization, information extraction, translation and categorization. Its coding capabilities include SQL generation, code explanations and translation between programming languages. It accepts and produces text, with language coverage that includes English, Arabic, Chinese, Japanese, Ukrainian and Persian.
The weights use Safetensors and work with Hugging Face Transformers. Cohere describes deployment on two GPUs, making this a model for teams with access to substantial GPU hardware. Its long context supports document-heavy tasks. Contextual and strict safety modes provide different restrictions on sensitive content.
The research weights carry the CC-BY-NC-4.0 license and Cohere Labs' Acceptable Use Policy. Access requires a Hugging Face account, acceptance of the terms and sharing contact information with Cohere.
