DeepSeek-V3.2

DeepSeek

Katalgu biss Piżijiet miftuħa

Sovranità u hosting

Katalgu biss

Dan il-mudell huwa katalogat għal referenza. Mhuwiex ospitat fuq LLM EU bħalissa, u l-ebda endpoint tal-ewwel parti ma jservih hawn. Talbiet lejh imorru direttament għand il-fornitur, taħt it-termini tal-fornitur stess.

Sommarju

DeepSeek-V3.2 is a 685B mixture-of-experts model released under the MIT licence, with a configuration maximum of 163,840 tokens. Its main architectural change is DeepSeek Sparse Attention, which the vendor describes as reducing the computational cost of long-context inference. It is a text-only model and needs a multi-node deployment at full precision.

Speċifikazzjonijiet

Fornitur
DeepSeek (CN)
Tieqa tal-kuntest
164k
Parametri
685.36B
Modalitajiet
text
Liċenza
mit · licence
Piżijiet miftuħa
yes
Data tal-ħruġ
01 Diċ 2025
Kompiti
reasoning, coding, summarization, rag

Lingwi

Qawwi: en, zh

Adekwat: de, fr, es, it, pt, ru, ja, ko, ar

Lingwi

Qawwi

  • en
  • zh

Adekwat

  • de
  • fr
  • es
  • it
  • pt
  • ru
  • ja
  • ko
  • ar

Mhux verifikat: il-fornitur ma dokumentax din id-dikjarazzjoni. — a language is listed as strong only where a source claims it; the card links the evidence where one exists.

Meta tużaha

Saħħiet

  • MIT licence
  • sparse attention aimed at lower long-context inference cost
  • 163,840-token configuration maximum

Limiti

  • 685B parameters: full-precision serving needs several nodes
  • vendor efficiency claims are not independently reproduced here
  • no evaluation has been run by LLM EU