- What is the primary use case for Command R7B?
- Command R7B is engineered specifically for Retrieval-Augmented Generation (RAG) and tool-use applications within enterprise environments, such as enterprise search, customer support bots, and AI agents interacting with APIs.
- How does Command R7B handle citations and accuracy?
- The model uses a 'grounded generation' approach that forces it to cite specific snippets of retrieved text for every claim, minimizing hallucinations. It is trained on 'RAG-first' data curation to accurately handle long-context retrieval and ignore irrelevant noise.
- What are the pricing details for Command R7B?
- The API costs $0.15 per 1M input tokens and $0.60 per 1M output tokens. Self-hosting is free under the Community License for companies with less than $50M in revenue, and enterprise customers can access custom volume-based pricing.
- What are the limitations of Command R7B?
- As a 7-billion parameter model, it has lower 'world knowledge' compared to larger models and may struggle with highly abstract creative writing tasks. It also requires careful prompt structuring for optimal citation results.