Title: DeepSeek-V3.2 %E2%80%94 deepseek-v3-2
Canonical: https://llmeu.com/sl/models/deepseek-v3-2/
Markdown mirror: https://llmeu.com/sl/models/deepseek-v3-2.md
Machine-readable: https://llmeu.com/llms.txt (index) - https://llmeu.com/llms-full.txt (full catalogue) - https://llmeu.com/llms-catalogue.json
Provider: DeepSeek
Licence: mit
Hosting: catalogue only %E2%80%94 not hosted on LLM EU yet
Sources: https://huggingface.co/deepseek-ai/DeepSeek-V3.2

# DeepSeek-V3.2

[DeepSeek](/sl/providers/deepseek/)

Nazadnje pregledano: 14. okt. 2026

## Suverenost in gostovanje

Ta model je v katalogu za referenco. Na LLM EU %C5%A1e ni gostovan in noben lastni endpoint ga tukaj ne stre%C5%BEe. Zahteve zanj gredo neposredno k ponudniku, pod lastnimi pogoji tega ponudnika.

## Povzetek

DeepSeek-V3.2 je model me%C5%A1anice strokovnjakov s 685B parametri, izdan pod licenco MIT, z najve%C4%8Djo konfiguracijo 163.840 %C5%BEetonov. Njegova glavna arhitekturna sprememba je DeepSeek Sparse Attention, ki jo prodajalec opisuje kot zmanj%C5%A1anje ra%C4%8Dunske cene sklepanja z dolgim kontekstom. Je model samo za besedilo in potrebuje ve%C4%8Dvozli%C5%A1%C4%8Dno uvedbo z natan%C4%8Dnostjo polne natan%C4%8Dnosti.

## Specifikacije

- Ponudnik
  DeepSeek (CN)
- Kontekstno okno
  164k
- Parametri
  685.36B
- Modalnosti
  text
- Licenca
  mit %C2%B7 [licence](https://opensource.org/license/mit)
- Odprte ute%C5%BEi
  yes
- Datum izdaje
  01. dec. 2025
- Naloge
  [Razmi%C5%A1ljanje](/sl/tasks/reasoning/), [Kodiranje](/sl/tasks/coding/), [Povzemanje](/sl/tasks/summarization/), [Pridobivanje (RAG)](/sl/tasks/rag/)

## Jeziki

**Mo%C4%8Dno:** en, zh

**Zadostno:** de, fr, es, it, pt, ru, ja, ko, ar

#### Mo%C4%8Dno

- en
- zh

#### Zadostno

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

Ni preverjeno: ponudnik te trditve ni dokumentiral. %E2%80%94 a language is listed as strong only where a source claims it; the card links the evidence where one exists.

## Kdaj ga uporabiti

#### Prednosti

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

#### Omejitve

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

## Pogosto zastavljena vpra%C5%A1anja

### Is DeepSeek-V3.2 served on LLM EU?

It has no endpoint in the catalogue, so it is catalogue only %E2%80%94 not hosted on LLM EU yet, with no region and no retention to report. A card whose status is "listed" is a catalogue entry, not a running deployment.

### Under what licence is DeepSeek-V3.2 released?

DeepSeek-V3.2 is recorded under mit. The recorded licence text is https://opensource.org/license/mit. The weights are downloadable (open_weights is true). LLM EU republishes the licence id and the link and gives no legal advice; the licence text governs.

### How large is DeepSeek-V3.2, and what context window does it have?

DeepSeek-V3.2 is recorded at 685B (safetensors total 685,355,329,792) (source: https://huggingface.co/api/models/deepseek-ai/DeepSeek-V3.2), with a context window of 163840 tokens (source: https://huggingface.co/deepseek-ai/DeepSeek-V3.2/blob/main/config.json).

### Cene

Ta model je v katalogu za referenco. Na LLM EU %C5%A1e ni gostovan in noben lastni endpoint ga tukaj ne stre%C5%BEe. Zahteve zanj gredo neposredno k ponudniku, pod lastnimi pogoji tega ponudnika.

### Status preverjanja

Kadar polje nima vira, kartica to prika%C5%BEe kot nepreverjeno, namesto da bi ga izpolnila. Prazno polje je informacija.

LLM EU Verified pomeni dokumentirano, ne pravno certificirano. To ni ocena skladnosti z aktom o umetni inteligenci in LLM EU ni organ EU.

### Razli%C4%8Dice in variante

- catalogue %C2%B7 Samo v katalogu %C2%B7 [weights](https://huggingface.co/deepseek-ai/DeepSeek-V3.2)

### Primer zahteve

```
curl https://api.llmeu.com/v1/chat/completions \
  -H "Authorization: Bearer $LLMEU_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"deepseek-v3-2","messages":[{"role":"user","content":"Hello"}]}'
```

Za ta model ni name%C5%A1%C4%8Denega peskovnika: nobena kon%C4%8Dna to%C4%8Dka ga ne stre%C5%BEe, zato zahteva ne bi imela kam iti.

### Benchmarki

Ni objavljene ocene. To je vrzel, ne skrit rezultat.

### Viri

- [huggingface.co](https://huggingface.co/deepseek-ai/DeepSeek-V3.2)

### Dnevnik sprememb kartice

- 01. dec. 2025 %E2%80%94 Datum izdaje
- 14. okt. 2026 %E2%80%94 Nazadnje pregledano

Nazadnje pregledano: 14. okt. 2026