Side-by-side comparison of Command R+ (Cohere · Canada) and DeepSeek R1 (DeepSeek · China) for self-hosted deployment of the open-weight model. Command R+ is rated conditional; DeepSeek R1 is conditional. They part ways on licence: Command R+ is "CC-BY-NC 4.0", DeepSeek R1 is "MIT".
| Field | ||
|---|---|---|
| Summary | ||
| Verdict | Conditional Enterprise-focused 104B model, strong at RAG and multilingual tool use. Weights are released under CC-BY-NC 4.0 — non-commercial only. Commercial deployment goes through Cohere's API (see Cohere API entry). | Conditional Frontier reasoning model at o1-class performance. MIT licence makes weights legally clean. Same Chinese-origin alignment/supply-chain considerations as DeepSeek V3. Distilled Qwen/Llama versions inherit their base licence. |
| Last reviewed | 2026-04-15 | 2026-04-15 |
| Open-weight | ||
| Licence | CC-BY-NC 4.0 | MIT |
| Commercial use | API only | Yes |
| Training data | Partial | Undisclosed |
| Origin | Canada | China |
| Performance & pricing? | ||
| Quality index | 8/100 | 27/100 |
| Speed | — | — |
| Blended price | $6.00/M | $2.36/M |
| Context window | — | — |
| Evidence | ||
| Sources | ||
No overlapping sources between the two entries.