architectures · deepseek-ai
Multi-head Latent Attention with fine-grained MoE and auxiliary-loss-free balancing. Strong loss-per-FLOP at scale.
| Weight file SHA-256 | de539966b9f54c2cdd8daa1891dde9d0c30a5ef629a3a04e2db0a83688bddc65 |
|---|
| Publisher | deepseek-ai |
|---|---|
| Origin | Recorded as a covered nation under 10 U.S.C. § 4872(f) — the PRC, Russia, Iran or North Korea. The registry did not record which. |
| Licence | MIT |
| Pulls recorded here | 64,001 |
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.