LANGUAGE MODEL Cohere Last updated:

Cohere Command R+

RAG-Optimized Open Frontier

Cohere's flagship open-weight model, released April 2024. Differentiated by being explicitly trained for retrieval- augmented generation (RAG) workflows — given retrieved documents in context, it integrates them more accurately and cites them more reliably than generic LLMs. The default model behind much enterprise RAG infrastructure in 2024-25.

Intelligence
Below avg
Cost
High
$3.00 in / $15.00 out
Context
128K
Up to 128,000 tokens
How are Intelligence, Speed & Cost bucketed?
Intelligence and Speed buckets are percentile ranks on Artificial Analysis. Cost buckets are fixed dollar thresholds keyed off output-token price ($/M out).
Intelligence
  • Top 1%≤ 1%
  • Top 5%≤ 5%
  • Top 10%≤ 10%
  • Good≤ 25%
  • Medium≤ 50%
  • Below avg> 50%
Speed
  • Top 1%≥ 345 tok/s
  • Top 5%≥ 237 tok/s
  • Top 10%≥ 196 tok/s
  • Good≥ 146 tok/s
  • Medium≥ 90 tok/s
  • Slow< 90 tok/s
Cost
  • Freeopen weights · self-host
  • Low< $1 / M out
  • Moderate$1–5 / M out
  • High≥ $5 / M out

Why it matters

Command R+ is the proof that "specialized for enterprise" can be a viable LLM-business positioning distinct from "race to the frontier." The continued existence of Cohere as a meaningful enterprise vendor (vs being absorbed by a hyperscaler) validates the segment.

Core Capabilities

Long Documents
Handles entire codebases, books, and multi-doc RAG.
Generative
Produces images, video, audio, or other media.
Agent Workflows
Built for tool use and autonomous tasks.

Context Window

128k tokens
≈ 98 pages
4k Chat 聊天
32k Long docs 长文档
128k This model 本模型
400k Multi-doc 多文档
1M Codebase 整个代码库
10M

Availability

API
Not available
Product / App
Not available
Open Source
Released
Enterprise

Pricing Model

Free / self-host
Open weights — pay only for compute.
Self-host

Capability / Performance

Where this model sits relative to the middle 60% of models in the tree. All scores are 0–10 (higher is better).

Lower 20% Upper 80% This model
Reasoning
AA Intelligence Index · scaled to 10
1.7
5.6
1.2
Coding
SciCode · scaled to 10
1.8
4.3
1.2
Context / memory
Context window size · log-scaled
6.0
9.0
6.0
Cost efficiency
Input price ($/M tokens) · cheaper scores higher
6.2
10.0
5.1
Lower 20% 20th percentile — 20% of models score below this This model Where the current model lands Upper 80% 80th percentile — only 20% of models score above this Percentile boundaries are computed across every model in the tree that reports the underlying benchmark for each capability.

What it feels like

  • Industry-leading RAG with in-line citations baked into the response — Cohere's signature differentiator
  • 104B params, 88.2% MMLU at release — competitive with GPT-4 Turbo on enterprise tasks
  • Beats Claude 3 Sonnet, Mistral Large, and GPT-4 Turbo on Cohere's internal RAG / tool-use benchmarks
  • Strong on Berkeley Function Calling and ToolTalk Hard — designed for multi-step agent flows
  • Multilingual coverage in 10 major business languages (EN, FR, DE, ES, IT, JA, KO, AR, PT, ZH)
  • Optimised for the enterprise sales motion — RAG citations matter more here than benchmark headlines

Best use cases

  • Enterprise RAG where citation accuracy and hallucination reduction matter most
  • Multi-step tool-use agents in regulated industries
  • Multilingual customer-facing assistants beyond English
  • Workflows needing predictable, citation-backed enterprise output

Tools to try

Not ideal for

  • Frontier-leaderboard reasoning by 2025 — newer Cohere Command-A and competitors lead
  • Consumer / chatbot use cases where personality and creativity matter
  • Coding-agent benchmarks (Claude Sonnet / GPT-5 stronger here)