Command R+
Last updated: August 10, 2026
Specs
| Vendor | Cohere |
|---|
| Released | April 4, 2024 |
|---|
| Context window | 128K tokens |
|---|
| Input | $2.5 / M tokens |
|---|
| Output | $10 / M tokens |
|---|
Benchmarks
| mmlu | 75.7 |
|---|
| gpqa | 42 |
|---|
| humanEval | 71.5 |
|---|
| arenaElo | 1213 |
|---|
| sweBench | 20 |
|---|
| sweBenchPro | 12 |
|---|
| liveCodeBench | 35 |
|---|
| aiderPolyglot | 30 |
|---|
| terminalBench | 10 |
|---|
| mmluPro | 55 |
|---|
| hle | 3 |
|---|
| aime | 25 |
|---|
| math500 | 60 |
|---|
Strengths
- Built for enterprise RAG, citations, grounding, retrieval
- Strong tool use and multi-step workflows
- Data sovereignty options (private deployment on your VPC)
Weaknesses
- Benchmarks lag behind frontier models
- Smaller developer mindshare
- No native multimodal
Best for
- Enterprise knowledge bots with strict source citation
- RAG-heavy workflows
- Regulated industries (finance, healthcare)
Head-to-head comparisons
Compare with other AI models