Title: DeepSeek-R1 | LLM EU
Canonical: https://llmeu.com/da/models/deepseek-r1/
Markdown mirror: https://llmeu.com/da/models/deepseek-r1.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-R1 https://arxiv.org/abs/2501.12948

# DeepSeek-R1

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

Sidst gennemg%C3%A5et: 14. okt. 2026

## Suver%C3%A6nitet og hosting

Denne model er katalogiseret til reference. Den er endnu ikke hostet p%C3%A5 LLM EU, og ingen f%C3%B8rsteparts-endpoint betjener den her. Anmodninger til den g%C3%A5r direkte til udbyderen under den p%C3%A5g%C3%A6ldende udbyders egne vilk%C3%A5r.

## Resum%C3%A9

DeepSeek-R1 er den r%C3%A6sonnementsmodel, der gjorde open-weight chain-of-thought tr%C3%A6ning bredt reproducerbar; den har 671B totale og 37B aktiverede parametre og er udgivet under MIT-licensen. Modelkortet dokumenterer en kontekstl%C3%A6ngde p%C3%A5 128K og rapporterer 90,8 p%C3%A5 MMLU, 84,0 p%C3%A5 MMLU-Pro og 79,8 p%C3%A5 AIME 2024 for R1. Det er en tekst-kun model og er blevet erstattet til de fleste opgaver af DeepSeek-V3.2 og V4.

## Specifikationer

- Udbyder
  DeepSeek (CN)
- Kontekstvindue
  131k
- Parametre
  684.49B
- Modaliteter
  text
- Licens
  mit %C2%B7 [licence](https://opensource.org/license/mit)
- %C3%85bne v%C3%A6gte
  yes
- Udgivelsesdato
  20. jan. 2025
- Opgaver
  [R%C3%A6sonnement](/da/tasks/reasoning/), [Kodning](/da/tasks/coding/), [Resum%C3%A9](/da/tasks/summarization/)

## Sprog

**St%C3%A6rk:** en, zh

**Tilstr%C3%A6kkelig:** de, fr, es, it, pt, ru, ja, ko

#### St%C3%A6rk

- en
- zh

#### Tilstr%C3%A6kkelig

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

Ikke verificeret: udbyderen har ikke dokumenteret denne p%C3%A5stand. %E2%80%94 a language is listed as strong only where a source claims it; the card links the evidence where one exists.

## Hvorn%C3%A5r man bruger den

#### Styrker

- MIT licence for a 671B-parameter reasoning model
- published reasoning traces in the base training recipe
- strong maths scores for a January 2025 release

#### Begr%C3%A6nsninger

- 671B parameters, so self-hosting is expensive
- reasoning output is verbose and raises token cost
- superseded by DeepSeek-V3.2 and V4-Flash for most workloads

## Ofte stillede sp%C3%B8rgsm%C3%A5l

### Is DeepSeek-R1 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-R1 released?

DeepSeek-R1 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-R1, and what context window does it have?

DeepSeek-R1 is recorded at 671B total, 37B activated (safetensors total 684,489,845,504) (source: https://huggingface.co/deepseek-ai/DeepSeek-R1), with a context window of 131072 tokens (source: https://huggingface.co/deepseek-ai/DeepSeek-R1).

### Priser

Denne model er katalogiseret til reference. Den er endnu ikke hostet p%C3%A5 LLM EU, og ingen f%C3%B8rsteparts-endpoint betjener den her. Anmodninger til den g%C3%A5r direkte til udbyderen under den p%C3%A5g%C3%A6ldende udbyders egne vilk%C3%A5r.

### Verifikationsstatus

Hvor et felt ikke har nogen kilde, viser kortet det som uverificeret i stedet for at udfylde det. Et tomt felt er information.

LLM EU Verified betyder dokumenteret, ikke juridisk certificeret. Det er ikke en overensstemmelsesvurdering efter AI Act, og LLM EU er ikke et EU-organ.

### Versioner og varianter

- catalogue %C2%B7 Kun katalog %C2%B7 [weights](https://huggingface.co/deepseek-ai/DeepSeek-R1)

### Eksempel p%C3%A5 anmodning

```
curl https://api.llmeu.com/v1/chat/completions \
  -H "Authorization: Bearer $LLMEU_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"deepseek-r1","messages":[{"role":"user","content":"Hello"}]}'
```

Ingen playground er udrullet til denne model: intet endpoint betjener den, s%C3%A5 en anmodning ville ikke have noget sted at g%C3%A5 hen.

### Benchmarks

- MMLU (Pass@1): **90.8 (vendor-reported)** [source](https://huggingface.co/deepseek-ai/DeepSeek-R1)
- MMLU-Pro (EM): **84.0 (vendor-reported)** [source](https://huggingface.co/deepseek-ai/DeepSeek-R1)
- GPQA-Diamond (Pass@1): **71.5 (vendor-reported)** [source](https://huggingface.co/deepseek-ai/DeepSeek-R1)
- AIME 2024 (Pass@1): **79.8 (vendor-reported)** [source](https://huggingface.co/deepseek-ai/DeepSeek-R1)
- LiveCodeBench (Pass@1-COT): **65.9 (vendor-reported)** [source](https://huggingface.co/deepseek-ai/DeepSeek-R1)

### Kilder

- [huggingface.co](https://huggingface.co/deepseek-ai/DeepSeek-R1)
- [arxiv.org](https://arxiv.org/abs/2501.12948)

### Kortets %C3%A6ndringslog

- 20. jan. 2025 %E2%80%94 Udgivelsesdato
- 14. okt. 2026 %E2%80%94 Sidst gennemg%C3%A5et

Sidst gennemg%C3%A5et: 14. okt. 2026

### Opgaver

- [R%C3%A6sonnement](/da/tasks/reasoning/)