DeepSeek

R1 / V3 671B-A37B

reasoningMoE

684,5B

Parameters · 37B active

381,1 Go

Q4_K_M file

≥ 382,6 Go

Recommended VRAM

163 840

Max context

Not measured yet

No card has run the TokenBench protocol on this model yet. The current protocol covers a selection of models; benchmarking any catalog model from the app is coming soon.

Measure my card →

Which cards should it fit on?

Cards on the site whose video memory covers the recommended 382,6 Go (file + 4,000-token context). An estimate, not a measurement. Click a card for the detailed estimate.

Technical sheet

Published
2025-05-28
Architecture
61 layers · 128 KV heads · dimension 56
Attention
Latent (MLA): the context cache is compressed — much lighter than the standard formula.
GGUF file
bartowski/deepseek-ai_DeepSeek-R1-0528-GGUF · 11 files Download (381,1 Go)

Public Hugging Face data (official config.json, GGUF repo). The context cache only counts full-attention layers.