c4ai-command-r-plus — benchmark results

Cohere's 104B open-weights (CC-BY-NC) model optimized for enterprise RAG with citations and multi-step tool use, with a 128K context. Provider: Cohere. Released 2024-04-03. Access: Open.

Unified ELO 1465 ± 21, rank #931 of 1776 rated models, from 42 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
Open LLM Leaderboard - MuSR20.42Score95.5
HumanLikeness - Sound-179.19Humanlike Score (%)94.7
LogicKor - Coding9.57Score (0-10)93.4
Open LLM Leaderboard - IFEval76.64Score92.4
HumanLikeness - Word-165.8Humanlike Score (%)89.5
LogicKor - Grammar7.78Score (0-10)88.9
LogicKor - Multi-Turn8.09Score (0-10)86.7
LogicKor - Understanding9.28Score (0-10)84.3
LogicKor7.89Score (0-10)84.1
LogicKor - Reasoning7.5Score (0-10)82.7
LogicKor - Writing9.14Score (0-10)82.5
Open LLM Leaderboard - BBH39.92Score79.1

Interactive version: theaggregate.ai/model?slug=c4ai-command-r-plus · How the rankings work · Data refreshed daily, snapshot 2026-07-22.