architectures · cohere
Sparse mixture-of-experts transformer behind the North family: 128 experts with 8 routed per token, grouped-query attention with a 4-way key/value grouping, and sliding-window attention. Declared as model_type `cohere2_moe`.
| Publisher | cohere |
|---|---|
| Origin | Canada |
| Licence | CC-BY-NC-4.0 |
Registry record as of 2026-08-14. Identity, licence and parameter count are properties of a release and do not move; tier can, and the verify link above re-checks it against the live registry rather than this page.