github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-08 00:51:34 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #13297] Downloading from HuggingFace asks me for a token #70844

New Issue
Open
opened 2026-05-04 23:10:05 -05:00 by GiteaMirror · 22 comments
GiteaMirror commented 2026-05-04 23:10:05 -05:00
Owner
Copy Link

Originally created by @vk2r on GitHub (Dec 2, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/13297

What is the issue?

I don't know what happened, but I got the Following error (I don't know why, since I've only downloaded one model in the last six days).

Image

It also shows an IP address that doesn't correspond to mine (or any other known IP address). I also didn't know that a token was needed for HF.

I don't know what's causing this.

OS

Windows 11

GPU

NVidia 3090

Ollama version

0.13.0

Originally created by @vk2r on GitHub (Dec 2, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/13297 ### What is the issue? I don't know what happened, but I got the Following error (I don't know why, since I've only downloaded one model in the last six days). <img width="1247" height="86" alt="Image" src="https://github.com/user-attachments/assets/b4e962f1-3930-4fe2-847c-9c474ae36c8e" /> It also shows an IP address that doesn't correspond to mine (or any other known IP address). I also didn't know that a token was needed for HF. I don't know what's causing this. ### OS Windows 11 ### GPU NVidia 3090 ### Ollama version 0.13.0
GiteaMirror added the bug label 2026-05-04 23:10:05 -05:00
GiteaMirror commented 2026-05-04 23:10:13 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Dec 2, 2025):

Obscuring the information that triggered the error (the IP address) makes it hard to identify the problem. At a guess, I'd say it's an internal IP address (does it start with 10?) and that what you are experiencing is a temporary internal issue. The model downloads fine for me.

<!-- gh-comment-id:3602768929 --> @rick-github commented on GitHub (Dec 2, 2025): Obscuring the information that triggered the error (the IP address) makes it hard to identify the problem. At a guess, I'd say it's an internal IP address (does it start with 10?) and that what you are experiencing is a temporary internal issue. The model downloads fine for me.
GiteaMirror commented 2026-05-04 23:10:18 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Dec 2, 2025):

Note this model is available in the ollama library: ministral-3:8b-instruct-q8_0

<!-- gh-comment-id:3602833923 --> @rick-github commented on GitHub (Dec 2, 2025): Note this model is available in the ollama library: [ministral-3:8b-instruct-q8_0](https://ollama.com/library/ministral-3:8b-instruct-q8_0)
GiteaMirror commented 2026-05-04 23:10:21 -05:00
Author
Owner
Copy Link

@pdevine commented on GitHub (Dec 2, 2025):

I haven't looked at the KVs/tensors for the unsloth version, but it may have problems initially running in Ollama if they are named something different. Also, I don't know if the unsloth version will include the vision component.

<!-- gh-comment-id:3604150403 --> @pdevine commented on GitHub (Dec 2, 2025): I haven't looked at the KVs/tensors for the unsloth version, but it may have problems initially running in Ollama if they are named something different. Also, I don't know if the unsloth version will include the vision component.
GiteaMirror commented 2026-05-04 23:10:23 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Dec 2, 2025):

The image projector is a separate GGUF.

<!-- gh-comment-id:3604157616 --> @rick-github commented on GitHub (Dec 2, 2025): The image projector is a separate GGUF.
GiteaMirror commented 2026-05-04 23:10:25 -05:00
Author
Owner
Copy Link

@pdevine commented on GitHub (Dec 2, 2025):

As to the error, I'm guessing it was happening before the repo was public? I used Ollama's version to transcribe the text of the error:

>>> transcribe the text in this image: /Users/pdevine/Documents/521422840-b4e962f1-3930-4fe2-847c-9c474ae36c8e.png
Added image '/Users/pdevine/Documents/521422840-b4e962f1-3930-4fe2-847c-9c474ae36c8e.png'
The text in the image reads as follows:

ollama pull hf.co/unsloth/Ministral-3-8B-Instruct-2512-GGUF:Q8_K_XL
pulling manifest
Error: pull model manifest: 429: {"pass": "We had to rate limit your IP.
existing account, and make sure you're using the IP that you registered
to continue our service, create a HF account or login to your
existing account."}


>>> Send a message (/? for help)

I don't think it liked the blacked out part, and I'm not sure why it said "pass" instead of "error". :-D

<!-- gh-comment-id:3604166253 --> @pdevine commented on GitHub (Dec 2, 2025): As to the error, I'm guessing it was happening before the repo was public? I used Ollama's version to transcribe the text of the error: ```ollama run ministral-3 >>> transcribe the text in this image: /Users/pdevine/Documents/521422840-b4e962f1-3930-4fe2-847c-9c474ae36c8e.png Added image '/Users/pdevine/Documents/521422840-b4e962f1-3930-4fe2-847c-9c474ae36c8e.png' The text in the image reads as follows: ``` ollama pull hf.co/unsloth/Ministral-3-8B-Instruct-2512-GGUF:Q8_K_XL pulling manifest Error: pull model manifest: 429: {"pass": "We had to rate limit your IP. existing account, and make sure you're using the IP that you registered to continue our service, create a HF account or login to your existing account."} ``` >>> Send a message (/? for help) ``` I don't think it liked the blacked out part, and I'm not sure why it said "pass" instead of "error". :-D
GiteaMirror commented 2026-05-04 23:10:29 -05:00
Author
Owner
Copy Link

@vk2r commented on GitHub (Dec 3, 2025):

With the new version, this no longer happens. I will close this issue.

Thank you all!

<!-- gh-comment-id:3604848433 --> @vk2r commented on GitHub (Dec 3, 2025): With the new version, this no longer happens. I will close this issue. Thank you all!
GiteaMirror commented 2026-05-04 23:10:31 -05:00
Author
Owner
Copy Link

@bplrdev commented on GitHub (Jan 17, 2026):

Same issue here with the latest version (0.14.2). Reinstalling doesn't help.

pulling manifest
Error: pull model manifest: 429: {"error":"We had to rate limit your IP (10.0.113.87). To continue using our service, create a HF account or login to your existing account, and make sure you pass a HF_TOKEN if you're using the API."}
<!-- gh-comment-id:3764054976 --> @bplrdev commented on GitHub (Jan 17, 2026): Same issue here with the latest version (0.14.2). Reinstalling doesn't help. ``` pulling manifest Error: pull model manifest: 429: {"error":"We had to rate limit your IP (10.0.113.87). To continue using our service, create a HF account or login to your existing account, and make sure you pass a HF_TOKEN if you're using the API."} ```
GiteaMirror commented 2026-05-04 23:10:33 -05:00
Author
Owner
Copy Link

@rdeforest commented on GitHub (Jan 17, 2026):

I'm seeing this (rate limit, HF_TOKEN) with ollama pull hf.co/BeaverAI/Cydonia-24B-v4zm-GGUF:Q6_K. I'm downloading the .gguf manually for now but wanted to comment that I'm available for troubleshooting if it would be helpful.

Edit to add: the problem no longer happens. It appears to have been something transient, suggesting on the HF side. Still, if there's any way to improve the user experience, that would be great.

Edit 2: it pulled the manifest and the first file, then failed on the next file (429). So I dug some more. I don't know if ollama uses HF_TOKEN, but I did an export HF_TOKEN=$(cat ~/.cache/huggingface/token) and it didn't help at all.

I'm now using hf download instead and having no issues. Is this an hf.co issue or an ollama issue?

<!-- gh-comment-id:3764160339 --> @rdeforest commented on GitHub (Jan 17, 2026): I'm seeing this (rate limit, HF_TOKEN) with `ollama pull hf.co/BeaverAI/Cydonia-24B-v4zm-GGUF:Q6_K`. I'm downloading the .gguf manually for now but wanted to comment that I'm available for troubleshooting if it would be helpful. Edit to add: the problem no longer happens. It appears to have been something transient, suggesting on the HF side. Still, if there's any way to improve the user experience, that would be great. Edit 2: it pulled the manifest and the first file, then failed on the next file (429). So I dug some more. I don't know if ollama uses HF_TOKEN, but I did an `export HF_TOKEN=$(cat ~/.cache/huggingface/token)` and it didn't help at all. I'm now using `hf download` instead and having no issues. Is this an hf.co issue or an ollama issue?
GiteaMirror commented 2026-05-04 23:10:35 -05:00
Author
Owner
Copy Link

@Wijt commented on GitHub (Jan 18, 2026):

I'm encountering the same issue with the latest version.

Image

It seems the current implementation doesn't utilize the token even if it's exported. I'm considering submitting a PR to pass the HF_TOKEN correctly during the download process. Would this be the right approach? @vk2r

<!-- gh-comment-id:3765705019 --> @Wijt commented on GitHub (Jan 18, 2026): I'm encountering the same issue with the latest version. <img width="1487" height="116" alt="Image" src="https://github.com/user-attachments/assets/73dfa543-d1e9-46b6-8547-263fc551abfb" /> It seems the current implementation doesn't utilize the token even if it's exported. I'm considering submitting a PR to pass the `HF_TOKEN` correctly during the download process. Would this be the right approach? @vk2r
GiteaMirror commented 2026-05-04 23:10:38 -05:00
Author
Owner
Copy Link

@ttait1 commented on GitHub (Jan 18, 2026):

I'm seeing the same error too:
Error: pull model manifest: 429: {"error":"We had to rate limit your IP (10.0.113.87). To continue using our service, create a HF account or login to your existing account, and make sure you pass a HF_TOKEN if you're using the API."}

I don't even have a 10.x.x.x ip address in my network. Google suggests it must be cloudflare proxy ip being shared. I tried doing "HF_TOKEN=hf_xxxxxxxxxx" ollama pull hf.co/modelname" with no change.

<!-- gh-comment-id:3765740450 --> @ttait1 commented on GitHub (Jan 18, 2026): I'm seeing the same error too: Error: pull model manifest: 429: {"error":"We had to rate limit your IP (10.0.113.87). To continue using our service, create a HF account or login to your existing account, and make sure you pass a HF_TOKEN if you're using the API."} I don't even have a 10.x.x.x ip address in my network. Google suggests it must be cloudflare proxy ip being shared. I tried doing "HF_TOKEN=hf_xxxxxxxxxx" ollama pull hf.co/modelname" with no change.
GiteaMirror commented 2026-05-04 23:10:40 -05:00
Author
Owner
Copy Link

@Mikec78660 commented on GitHub (Jan 18, 2026):

Seeing the same. Put the export HF token option in my ollama.service file but still doesn't work. I tried a few times and each time it told it was different 10.x.x.x ip that was rate limited. Weird.

<!-- gh-comment-id:3765756340 --> @Mikec78660 commented on GitHub (Jan 18, 2026): Seeing the same. Put the export HF token option in my ollama.service file but still doesn't work. I tried a few times and each time it told it was different 10.x.x.x ip that was rate limited. Weird.
GiteaMirror commented 2026-05-04 23:10:42 -05:00
Author
Owner
Copy Link

@noes14155 commented on GitHub (Jan 18, 2026):

Same issue in docker installation 10.x.x.x ip rate limited

<!-- gh-comment-id:3765758830 --> @noes14155 commented on GitHub (Jan 18, 2026): Same issue in docker installation 10.x.x.x ip rate limited
GiteaMirror commented 2026-05-04 23:10:45 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Jan 18, 2026):

HF credentials are not currently supported in ollama. This looks like HF has implemented tighter rate limiting on their internal network and ollama's download methodology of multiple concurrent sessions is triggering the protection mechanism. Currently the only solution is to retry or to download the model via hf or browser.

<!-- gh-comment-id:3765759654 --> @rick-github commented on GitHub (Jan 18, 2026): HF credentials are not currently supported in ollama. This looks like HF has implemented tighter rate limiting on their internal network and ollama's download methodology of multiple concurrent sessions is triggering the protection mechanism. Currently the only solution is to retry or to download the model via `hf` or browser.
GiteaMirror commented 2026-05-04 23:10:50 -05:00
Author
Owner
Copy Link

@rdeforest commented on GitHub (Jan 18, 2026):

Ah, in that case, can we re-open this as a feature request?

<!-- gh-comment-id:3765760014 --> @rdeforest commented on GitHub (Jan 18, 2026): Ah, in that case, can we re-open this as a feature request?
GiteaMirror commented 2026-05-04 23:10:51 -05:00
Author
Owner
Copy Link

@ParthSareen commented on GitHub (Jan 18, 2026):

Taking a look into it folks!

<!-- gh-comment-id:3765773877 --> @ParthSareen commented on GitHub (Jan 18, 2026): Taking a look into it folks!
GiteaMirror commented 2026-05-04 23:10:54 -05:00
Author
Owner
Copy Link

@rdeforest commented on GitHub (Jan 18, 2026):

I use ollama to run models locally, so I don't need to subscribe to ollama.com, but I went ahead and did anyway to support development. Thanks guys!

<!-- gh-comment-id:3765810208 --> @rdeforest commented on GitHub (Jan 18, 2026): I use ollama to run models locally, so I don't need to subscribe to ollama.com, but I went ahead and did anyway to support development. Thanks guys!
GiteaMirror commented 2026-05-04 23:10:56 -05:00
Author
Owner
Copy Link

@SecKatie commented on GitHub (Jan 18, 2026):

Would it be possible to use the ollama ssh key to auth to HF for pulls? I am also seeing this issue and would prefer if it used the auth I already have instead of putting a token in my unit file.

<!-- gh-comment-id:3765822081 --> @SecKatie commented on GitHub (Jan 18, 2026): Would it be possible to use the ollama ssh key to auth to HF for pulls? I am also seeing this issue and would prefer if it used the auth I already have instead of putting a token in my unit file.
GiteaMirror commented 2026-05-04 23:10:58 -05:00
Author
Owner
Copy Link

@BarachielFallen commented on GitHub (Jan 19, 2026):

I was able to backdate my version of the ollama docker container to ollama:0.12.11 and the ollama pull command succeeded with the same huggingface file that failed with the ollama:latest version, version 0.14.1, and 0.14.2. So I can backdate to successfully pull models again with this version but then I have to recreate my docker container with the latest version in order to actually run those models since version 12 is missing the latest model supports. I am also seeing higher speeds that used to be capped at 10 MB/s previously:

ollama-webui# docker exec -it ollama ollama pull hf.co/DavidAU/Qwen3-128k-30B-A3B-NEO-MAX-Imatrix-gguf:IQ4_NL
pulling manifest
pulling 1f3b622d7da9: 75% ▕██████████████████████████████████████████████████████████████ ▏ 12 GB/ 17 GB 29 MB/s 2m27s

pulling 1f3b622d7da9: 100% ▕████████████████████████████████████████████████████████████████████████████████████▏ 17 GB

verifying sha256 digest
writing manifest
success

<!-- gh-comment-id:3766099719 --> @BarachielFallen commented on GitHub (Jan 19, 2026): I was able to backdate my version of the ollama docker container to ollama:0.12.11 and the ollama pull command succeeded with the same huggingface file that failed with the ollama:latest version, version 0.14.1, and 0.14.2. So I can backdate to successfully pull models again with this version but then I have to recreate my docker container with the latest version in order to actually run those models since version 12 is missing the latest model supports. I am also seeing higher speeds that used to be capped at 10 MB/s previously: ollama-webui# docker exec -it ollama ollama pull hf.co/DavidAU/Qwen3-128k-30B-A3B-NEO-MAX-Imatrix-gguf:IQ4_NL pulling manifest pulling 1f3b622d7da9: 75% ▕██████████████████████████████████████████████████████████████ ▏ 12 GB/ 17 GB 29 MB/s 2m27s pulling 1f3b622d7da9: 100% ▕████████████████████████████████████████████████████████████████████████████████████▏ 17 GB verifying sha256 digest writing manifest success
GiteaMirror commented 2026-05-04 23:11:02 -05:00
Author
Owner
Copy Link

@BarachielFallen commented on GitHub (Jan 19, 2026):

I was able to backdate my version of the ollama docker container to ollama:0.12.11 and the ollama pull command succeeded with the same huggingface file that failed with the ollama:latest version, version 0.14.1, and 0.14.2. So I can backdate to successfully pull models again with this version but then I have to recreate my docker container with the latest version in order to actually run those models since version 12 is missing the latest model supports. I am also seeing higher speeds that used to be capped at 10 MB/s previously:

ollama-webui# docker exec -it ollama ollama pull hf.co/DavidAU/Qwen3-128k-30B-A3B-NEO-MAX-Imatrix-gguf:IQ4_NL pulling manifest pulling 1f3b622d7da9: 75% ▕██████████████████████████████████████████████████████████████ ▏ 12 GB/ 17 GB 29 MB/s 2m27s

pulling 1f3b622d7da9: 100% ▕████████████████████████████████████████████████████████████████████████████████████▏ 17 GB

verifying sha256 digest writing manifest success

OK next attempt it fails:

docker exec -it ollama ollama pull hf.co/unsloth/Nemotron-3-Nano-30B-A3B-GGUF:IQ4_XS
pulling manifest
Error: pull model manifest: 429: {"error":"We had to rate limit your IP (10.x.x.190). To continue using our service, create a HF account or login to your existing account, and make sure you pass a HF_TOKEN if you're using the API."}

<!-- gh-comment-id:3766134714 --> @BarachielFallen commented on GitHub (Jan 19, 2026): > I was able to backdate my version of the ollama docker container to ollama:0.12.11 and the ollama pull command succeeded with the same huggingface file that failed with the ollama:latest version, version 0.14.1, and 0.14.2. So I can backdate to successfully pull models again with this version but then I have to recreate my docker container with the latest version in order to actually run those models since version 12 is missing the latest model supports. I am also seeing higher speeds that used to be capped at 10 MB/s previously: > > ollama-webui# docker exec -it ollama ollama pull hf.co/DavidAU/Qwen3-128k-30B-A3B-NEO-MAX-Imatrix-gguf:IQ4_NL pulling manifest pulling 1f3b622d7da9: 75% ▕██████████████████████████████████████████████████████████████ ▏ 12 GB/ 17 GB 29 MB/s 2m27s > > pulling 1f3b622d7da9: 100% ▕████████████████████████████████████████████████████████████████████████████████████▏ 17 GB > > verifying sha256 digest writing manifest success OK next attempt it fails: docker exec -it ollama ollama pull hf.co/unsloth/Nemotron-3-Nano-30B-A3B-GGUF:IQ4_XS pulling manifest Error: pull model manifest: 429: {"error":"We had to rate limit your IP (10.x.x.190). To continue using our service, create a HF account or login to your existing account, and make sure you pass a HF_TOKEN if you're using the API."}
GiteaMirror commented 2026-05-04 23:11:12 -05:00
Author
Owner
Copy Link

@rdeforest commented on GitHub (Jan 19, 2026):

classic hiesenbug :)

<!-- gh-comment-id:3766136266 --> @rdeforest commented on GitHub (Jan 19, 2026): classic hiesenbug :)
GiteaMirror commented 2026-05-04 23:11:17 -05:00
Author
Owner
Copy Link

@ParthSareen commented on GitHub (Jan 19, 2026):

WIP if any of you want to play with this in the meantime. Need to do some cleanup before merge + release! Sorry about this! https://github.com/ollama/ollama/pull/13769

<!-- gh-comment-id:3766137288 --> @ParthSareen commented on GitHub (Jan 19, 2026): WIP if any of you want to play with this in the meantime. Need to do some cleanup before merge + release! Sorry about this! https://github.com/ollama/ollama/pull/13769
GiteaMirror commented 2026-05-04 23:11:19 -05:00
Author
Owner
Copy Link

@BarachielFallen commented on GitHub (Jan 19, 2026):

OK now its working again on 0.12.11 and I am getting ridiculous speeds for my starlink connection:

/ollama-webui# docker exec -it ollama ollama pull hf.co/unsloth/Nemotron-3-Nano-30B-A3B-GGUF:IQ4_XS pulling manifest
pulling 0f2cbdc4f578: 91% ▕████████████████████████████████████████████████████████████████████████████ ▏ 16 GB/ 18 GB 47 MB/s 35s

<!-- gh-comment-id:3768164595 --> @BarachielFallen commented on GitHub (Jan 19, 2026): OK now its working again on 0.12.11 and I am getting ridiculous speeds for my starlink connection: /ollama-webui# docker exec -it ollama ollama pull hf.co/unsloth/Nemotron-3-Nano-30B-A3B-GGUF:IQ4_XS pulling manifest pulling 0f2cbdc4f578: 91% ▕████████████████████████████████████████████████████████████████████████████ ▏ 16 GB/ 18 GB 47 MB/s 35s
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
amd
api
app
bug
build
cli
cloud
compatibility
context-length
create
docker
documentation
embeddings
feature request
feedback wanted
good first issue
gpt-oss
gpu
harmony
help wanted
image
install
intel
js
launch
linux
macos
memory
mlx
model
needs more info
networking
nvidia
ollama.com
performance
pull-request
Mirrored from GitHub Pull Request
 python
question
registry
rendering
thinking
tools
top
vulkan
windows
wsl
No Label bug
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#70844
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?