Cohere is a Canadian AI company headquartered in Toronto. It builds large language models for businesses, including the Command family for text generation and Rerank and Embed models for search, and runs a research lab called Cohere Labs that released the Aya models.
- Founded
- 2019
- Type
- Private company
- Founders
- Aidan Gomez, Ivan Zhang and Nick Frosst
- AI lab
- Inference provider
- 17 hosted endpoints
| Model | Endpoint | Context | Input /M | Output /M | Cached input /M | Cache write /M | Uptime 1d |
|---|---|---|---|---|---|---|---|
 Command A | cohere | 256K | $2.50 | $10.00 | — | — | 100.00% |
| Command A+ | cohere/fp8 | 192K | $0.30 | $1.50 | $0.15 | — | 100.00% |
| Command R (08-2024) | cohere | 128K | $0.15 | $0.60 | — | — | 99.52% |
| Command R+ (08-2024) | cohere | 128K | $2.50 | $10.00 | — | — | 100.00% |
| Command R7B (12-2024) | cohere | 128K | $0.0375 | $0.15 | — | — | 100.00% |
 DeepSeek V4 Flash 0731 | cohere/fp8 | 1M | $0.14 | $0.28 | $0.07 | — | 99.93% |
| DeepSeek V4.1 Flash | cohere/fp8 | 1M | $0.195 | $0.66 | $0.003 | — | 100.00% |
| DeepSeek V4.1 Flash 1017 | cohere/fp8 | 1M | $0.15 | $0.50 | $0.003 | — | 100.00% |
| Embed 5 Fast | cohere | 128K | $0.08 | $0.00 | — | — | 100.00% |
| Embed 5 Pro | cohere | 128K | $0.12 | $0.00 | — | — | 100.00% |
 GLM 5.3 Flash | cohere/nvfp4 | 1M | $0.15 | $0.50 | $0.03 | — | 99.58% |
 MiMo-V2.6-Flash | cohere/fp8 | 1M | $0.14 | $0.28 | $0.003 | — | 99.44% |
| Rerank 4 Fast | cohere | 33K | $0.00 | $0.00 | — | — | 100.00% |
| Rerank 4 Pro | cohere | 33K | $0.00 | $0.00 | — | — | 100.00% |
| Rerank 5 Fast | cohere | 33K | $0.00 | $0.00 | — | — | 100.00% |
| Rerank 5 Pro | cohere | 33K | $0.00 | $0.00 | — | — | 100.00% |
| Rerank v3.5 | cohere | 4K | $0.00 | $0.00 | — | — | 100.00% |