[GH-ISSUE #8911] The DeepSeekR1:671B pulls down to a size of only 376G. #5778

Closed
opened 2026-04-12 17:07:11 -05:00 by GiteaMirror · 2 comments
Owner

Originally created by @SuperJunier666 on GitHub (Feb 7, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/8911

Isn't it 404G, I need to download it and then copy it before uploading it, so I'm wondering about that. Also is this size using the 4-bit quantization method?

Originally created by @SuperJunier666 on GitHub (Feb 7, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/8911 Isn't it 404G, I need to download it and then copy it before uploading it, so I'm wondering about that. Also is this size using the 4-bit quantization method?
GiteaMirror added the model label 2026-04-12 17:07:11 -05:00
Author
Owner

@SuperJunier666 commented on GitHub (Feb 7, 2025):

Image

Image

Image

<!-- gh-comment-id:2641840302 --> @SuperJunier666 commented on GitHub (Feb 7, 2025): ![Image](https://github.com/user-attachments/assets/5dfc8a0e-447f-4001-b87f-42fda064e69c) ![Image](https://github.com/user-attachments/assets/1f11917e-9435-470d-b960-0aa79d39c31f) ![Image](https://github.com/user-attachments/assets/906d888c-28cb-4356-a236-8d40de52f85f)
Author
Owner

@jmorganca commented on GitHub (Feb 7, 2025):

Yes, the 671b tag here is an alias for the 4-bit quantization: https://ollama.com/library/deepseek-r1/tags. A full precision fp16 model is available as well

<!-- gh-comment-id:2642000026 --> @jmorganca commented on GitHub (Feb 7, 2025): Yes, the `671b` tag here is an alias for the 4-bit quantization: https://ollama.com/library/deepseek-r1/tags. A full precision fp16 model is available as well
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#5778