Cohere: Command R7B (12-2024)Model26/100 via “instruction-following and prompt compliance”
Command R7B (12-2024) is a small, fast update of the Command R+ model, delivered in December 2024. It excels at RAG, tool use, agents, and similar tasks requiring complex reasoning...
Unique: Command R7B's instruction-following is optimized for RAG and tool-use contexts, where it must balance following user instructions with incorporating retrieved information and tool results
vs others: More reliable instruction compliance than GPT-3.5 Turbo on complex multi-constraint prompts, comparable to Claude 3 Opus but with lower latency