github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-07 00:22:43 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #14868] Extremely slow model download speed with ollama pull/ollama run, UI downloads and curl commands. #56100

New Issue
Closed
opened 2026-04-29 10:15:59 -05:00 by GiteaMirror · 6 comments
GiteaMirror commented 2026-04-29 10:15:59 -05:00
Owner
Copy Link

Originally created by @SingularityMan on GitHub (Mar 16, 2026).
Original GitHub issue: https://github.com/ollama/ollama/issues/14868

What is the issue?

Been running into this issue for a few days so I decided to upgrade today to 0.18.0 but the issue remains. For some reason, no matter which model I pull from ollama, the download speeds slow to a crawl, like sub-1KB/s download speeds.

My internet speed is in the 800mbps upload/download range, and Ollama was uninstalled/reinstalled and I even restarted my PC. Doesn't matter if I try the CLI, UI or even curl, downloading any model at all from Ollama's registry seems to be having issues for me to the point where the downloads even time out. And if I let it try to download it eventually returns this error in the CLI:

ollama pull qwen3.5:27b-q8_0
pulling manifest

----------------------------# ~30 minutes later-----------------------
Error: context canceled

I think it might be some networking issue but I'm using my home's network as usual. I've never run into this issue before. If there's something I can do on my end I am not sure what else to try. Of course, I'm not saying its Ollama's fault neither, but when I download a model on HF it downloads as fast as ever (although I can't run qwen3.5 GGUFs from HF...) but on Ollama all of a sudden it slows down too much.

Relevant log output

[GIN] 2026/03/15 - 21:57:16 | 200 |      1.5802ms |       127.0.0.1 | GET      "/api/tags"
time=2026-03-15T21:57:23.777-04:00 level=INFO source=download.go:376 msg="ab6e5325ec86 part 4 stalled; retrying. If this persists, press ctrl-c to exit, then 'ollama pull' to find a faster connection."
time=2026-03-15T21:57:38.704-04:00 level=INFO source=download.go:376 msg="ab6e5325ec86 part 2 stalled; retrying. If this persists, press ctrl-c to exit, then 'ollama pull' to find a faster connection."
[GIN] 2026/03/15 - 21:57:46 | 200 |      1.9995ms |       127.0.0.1 | GET      "/api/tags"
time=2026-03-15T21:57:50.824-04:00 level=INFO source=download.go:297 msg="ab6e5325ec86 part 2 attempt 0 failed: Get \"https://dd20bb891979d25aebc8bec07b2b3bbc.r2.cloudflarestorage.com/ollama/docker/registry/v2/blobs/sha256/ab/ab6e5325ec86fa9823354e219d684d654fbe97c9be41442b40dba09c4edf4af3/data?X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=66040c77ac1b787c3af820529859349a%2F20260316%2Fauto%2Fs3%2Faws4_request&X-Amz-Date=20260316T015505Z&X-Amz-Expires=86400&X-Amz-SignedHeaders=host&X-Amz-Signature=69447326e78a8c3eef332ff81021541651457abf43a0f451ba1ed1886790955e\": net/http: TLS handshake timeout, retrying in 1s"
time=2026-03-15T21:58:09.938-04:00 level=INFO source=download.go:297 msg="ab6e5325ec86 part 2 attempt 1 failed: Get \"https://dd20bb891979d25aebc8bec07b2b3bbc.r2.cloudflarestorage.com/ollama/docker/registry/v2/blobs/sha256/ab/ab6e5325ec86fa9823354e219d684d654fbe97c9be41442b40dba09c4edf4af3/data?X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=66040c77ac1b787c3af820529859349a%2F20260316%2Fauto%2Fs3%2Faws4_request&X-Amz-Date=20260316T015505Z&X-Amz-Expires=86400&X-Amz-SignedHeaders=host&X-Amz-Signature=69447326e78a8c3eef332ff81021541651457abf43a0f451ba1ed1886790955e\": net/http: TLS handshake timeout, retrying in 2s"
[GIN] 2026/03/15 - 21:58:16 | 200 |      1.5784ms |       127.0.0.1 | GET      "/api/tags"
time=2026-03-15T21:58:32.704-04:00 level=INFO source=download.go:376 msg="ab6e5325ec86 part 5 stalled; retrying. If this persists, press ctrl-c to exit, then 'ollama pull' to find a faster connection."
time=2026-03-15T21:58:44.834-04:00 level=INFO source=download.go:297 msg="ab6e5325ec86 part 5 attempt 0 failed: Get \"https://dd20bb891979d25aebc8bec07b2b3bbc.r2.cloudflarestorage.com/ollama/docker/registry/v2/blobs/sha256/ab/ab6e5325ec86fa9823354e219d684d654fbe97c9be41442b40dba09c4edf4af3/data?X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=66040c77ac1b787c3af820529859349a%2F20260316%2Fauto%2Fs3%2Faws4_request&X-Amz-Date=20260316T015505Z&X-Amz-Expires=86400&X-Amz-SignedHeaders=host&X-Amz-Signature=69447326e78a8c3eef332ff81021541651457abf43a0f451ba1ed1886790955e\": net/http: TLS handshake timeout, retrying in 1s"
[GIN] 2026/03/15 - 21:58:46 | 200 |      1.0547ms |       127.0.0.1 | GET      "/api/tags"

OS

Windows

GPU

Nvidia

CPU

AMD

Ollama version

0.18.0

Originally created by @SingularityMan on GitHub (Mar 16, 2026). Original GitHub issue: https://github.com/ollama/ollama/issues/14868 ### What is the issue? Been running into this issue for a few days so I decided to upgrade today to `0.18.0` but the issue remains. For some reason, no matter which model I pull from ollama, the download speeds slow to a crawl, like sub-1KB/s download speeds. My internet speed is in the 800mbps upload/download range, and Ollama was uninstalled/reinstalled and I even restarted my PC. Doesn't matter if I try the CLI, UI or even `curl`, downloading any model at all from Ollama's registry seems to be having issues for me to the point where the downloads even time out. And if I let it try to download it eventually returns this error in the CLI: ``` ollama pull qwen3.5:27b-q8_0 pulling manifest ----------------------------# ~30 minutes later----------------------- Error: context canceled ``` I think it might be some networking issue but I'm using my home's network as usual. I've never run into this issue before. If there's something I can do on my end I am not sure what else to try. Of course, I'm not saying its Ollama's fault neither, but when I download a model on HF it downloads as fast as ever (although I can't run qwen3.5 GGUFs from HF...) but on Ollama all of a sudden it slows down too much. ### Relevant log output ```shell [GIN] 2026/03/15 - 21:57:16 | 200 | 1.5802ms | 127.0.0.1 | GET "/api/tags" time=2026-03-15T21:57:23.777-04:00 level=INFO source=download.go:376 msg="ab6e5325ec86 part 4 stalled; retrying. If this persists, press ctrl-c to exit, then 'ollama pull' to find a faster connection." time=2026-03-15T21:57:38.704-04:00 level=INFO source=download.go:376 msg="ab6e5325ec86 part 2 stalled; retrying. If this persists, press ctrl-c to exit, then 'ollama pull' to find a faster connection." [GIN] 2026/03/15 - 21:57:46 | 200 | 1.9995ms | 127.0.0.1 | GET "/api/tags" time=2026-03-15T21:57:50.824-04:00 level=INFO source=download.go:297 msg="ab6e5325ec86 part 2 attempt 0 failed: Get \"https://dd20bb891979d25aebc8bec07b2b3bbc.r2.cloudflarestorage.com/ollama/docker/registry/v2/blobs/sha256/ab/ab6e5325ec86fa9823354e219d684d654fbe97c9be41442b40dba09c4edf4af3/data?X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=66040c77ac1b787c3af820529859349a%2F20260316%2Fauto%2Fs3%2Faws4_request&X-Amz-Date=20260316T015505Z&X-Amz-Expires=86400&X-Amz-SignedHeaders=host&X-Amz-Signature=69447326e78a8c3eef332ff81021541651457abf43a0f451ba1ed1886790955e\": net/http: TLS handshake timeout, retrying in 1s" time=2026-03-15T21:58:09.938-04:00 level=INFO source=download.go:297 msg="ab6e5325ec86 part 2 attempt 1 failed: Get \"https://dd20bb891979d25aebc8bec07b2b3bbc.r2.cloudflarestorage.com/ollama/docker/registry/v2/blobs/sha256/ab/ab6e5325ec86fa9823354e219d684d654fbe97c9be41442b40dba09c4edf4af3/data?X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=66040c77ac1b787c3af820529859349a%2F20260316%2Fauto%2Fs3%2Faws4_request&X-Amz-Date=20260316T015505Z&X-Amz-Expires=86400&X-Amz-SignedHeaders=host&X-Amz-Signature=69447326e78a8c3eef332ff81021541651457abf43a0f451ba1ed1886790955e\": net/http: TLS handshake timeout, retrying in 2s" [GIN] 2026/03/15 - 21:58:16 | 200 | 1.5784ms | 127.0.0.1 | GET "/api/tags" time=2026-03-15T21:58:32.704-04:00 level=INFO source=download.go:376 msg="ab6e5325ec86 part 5 stalled; retrying. If this persists, press ctrl-c to exit, then 'ollama pull' to find a faster connection." time=2026-03-15T21:58:44.834-04:00 level=INFO source=download.go:297 msg="ab6e5325ec86 part 5 attempt 0 failed: Get \"https://dd20bb891979d25aebc8bec07b2b3bbc.r2.cloudflarestorage.com/ollama/docker/registry/v2/blobs/sha256/ab/ab6e5325ec86fa9823354e219d684d654fbe97c9be41442b40dba09c4edf4af3/data?X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=66040c77ac1b787c3af820529859349a%2F20260316%2Fauto%2Fs3%2Faws4_request&X-Amz-Date=20260316T015505Z&X-Amz-Expires=86400&X-Amz-SignedHeaders=host&X-Amz-Signature=69447326e78a8c3eef332ff81021541651457abf43a0f451ba1ed1886790955e\": net/http: TLS handshake timeout, retrying in 1s" [GIN] 2026/03/15 - 21:58:46 | 200 | 1.0547ms | 127.0.0.1 | GET "/api/tags" ``` ### OS Windows ### GPU Nvidia ### CPU AMD ### Ollama version 0.18.0
GiteaMirror added the bug label 2026-04-29 10:15:59 -05:00
GiteaMirror commented 2026-04-29 10:16:03 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Mar 16, 2026):

Does the slowness persist if you use a VPN? FWIW I see no slowdown when pulling models in Europe.

<!-- gh-comment-id:4066109796 --> @rick-github commented on GitHub (Mar 16, 2026): Does the slowness persist if you use a VPN? FWIW I see no slowdown when pulling models in Europe.
GiteaMirror commented 2026-04-29 10:16:03 -05:00
Author
Owner
Copy Link

@SingularityMan commented on GitHub (Mar 16, 2026):

Does the slowness persist if you use a VPN? FWIW I see no slowdown when pulling models in Europe.

Yeah, as soon as I connected to a VPN in the midwest, it gave me extremely fast download speeds but I'm stationed in SW US so I don't know why that started slowing down my speeds with my regular connection. That's so weird.

EDIT: Wait, no, I take that back. Same download speeds. Gonna try a different region.

EDIT2: Ok, same thing switching to Europe. Super fast start then gradual slowdown. Not even %50 complete.

EDIT3: Same thing with Singapore. I see a pattern emerging here. As soon as I switch regions it gives me super fast download then 1 minute later the speeds drop to a KB/s speed. This is so weird.

pulling manifest
pulling ab6e5325ec86:  35% ▕█████████████████████████████████████████████████████                                                                                                    ▏ 5.5 GB/ 16 GB  368 KB/s   7h53m

[GIN] 2026/03/16 - 10:29:51 | 200 |            0s |       127.0.0.1 | HEAD     "/"
time=2026-03-16T10:29:52.314-04:00 level=INFO source=download.go:179 msg="downloading ab6e5325ec86 in 16 1 GB part(s)"
[GIN] 2026/03/16 - 10:31:45 | 200 |         1m53s |       127.0.0.1 | POST     "/api/pull"
[GIN] 2026/03/16 - 10:31:53 | 200 |            0s |       127.0.0.1 | HEAD     "/"
time=2026-03-16T10:31:55.009-04:00 level=INFO source=download.go:179 msg="downloading ab6e5325ec86 in 16 1 GB part(s)"
time=2026-03-16T10:32:47.232-04:00 level=INFO source=download.go:376 msg="ab6e5325ec86 part 9 stalled; retrying. If this persists, press ctrl-c to exit, then 'ollama pull' to find a faster connection."
<!-- gh-comment-id:4068093666 --> @SingularityMan commented on GitHub (Mar 16, 2026): > Does the slowness persist if you use a VPN? FWIW I see no slowdown when pulling models in Europe. Yeah, as soon as I connected to a VPN in the midwest, it gave me extremely fast download speeds but I'm stationed in SW US so I don't know why that started slowing down my speeds with my regular connection. That's so weird. EDIT: Wait, no, I take that back. Same download speeds. Gonna try a different region. EDIT2: Ok, same thing switching to Europe. Super fast start then gradual slowdown. Not even %50 complete. EDIT3: Same thing with Singapore. I see a pattern emerging here. As soon as I switch regions it gives me super fast download then 1 minute later the speeds drop to a KB/s speed. This is so weird. ``` pulling manifest pulling ab6e5325ec86: 35% ▕█████████████████████████████████████████████████████ ▏ 5.5 GB/ 16 GB 368 KB/s 7h53m [GIN] 2026/03/16 - 10:29:51 | 200 | 0s | 127.0.0.1 | HEAD "/" time=2026-03-16T10:29:52.314-04:00 level=INFO source=download.go:179 msg="downloading ab6e5325ec86 in 16 1 GB part(s)" [GIN] 2026/03/16 - 10:31:45 | 200 | 1m53s | 127.0.0.1 | POST "/api/pull" [GIN] 2026/03/16 - 10:31:53 | 200 | 0s | 127.0.0.1 | HEAD "/" time=2026-03-16T10:31:55.009-04:00 level=INFO source=download.go:179 msg="downloading ab6e5325ec86 in 16 1 GB part(s)" time=2026-03-16T10:32:47.232-04:00 level=INFO source=download.go:376 msg="ab6e5325ec86 part 9 stalled; retrying. If this persists, press ctrl-c to exit, then 'ollama pull' to find a faster connection." ```
GiteaMirror commented 2026-04-29 10:16:03 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Mar 16, 2026):

I just pulled qwen3.5:27b-q8_0 at an average of 556MB/s. It slowed down at the end as the faster chunks finished and the slower ones dominated the bandwidth, but it took less than a minute to download the model.

<!-- gh-comment-id:4068397601 --> @rick-github commented on GitHub (Mar 16, 2026): I just pulled qwen3.5:27b-q8_0 at an average of 556MB/s. It slowed down at the end as the faster chunks finished and the slower ones dominated the bandwidth, but it took less than a minute to download the model.
GiteaMirror commented 2026-04-29 10:16:04 -05:00
Author
Owner
Copy Link

@SingularityMan commented on GitHub (Mar 16, 2026):

I just pulled qwen3.5:27b-q8_0 at an average of 556MB/s. It slowed down at the end as the faster chunks finished and the slower ones dominated the bandwidth, but it took less than a minute to download the model.

K, so after hopping around different VPNs before finally using my own network again, It seems to have been fixed. I guess cloudflare throttles different IP addresses or something? I dunno, but I was basically throwing darts at the wall until it worked.

<!-- gh-comment-id:4069674151 --> @SingularityMan commented on GitHub (Mar 16, 2026): > I just pulled qwen3.5:27b-q8_0 at an average of 556MB/s. It slowed down at the end as the faster chunks finished and the slower ones dominated the bandwidth, but it took less than a minute to download the model. K, so after hopping around different VPNs before finally using my own network again, It seems to have been fixed. I guess cloudflare throttles different IP addresses or something? I dunno, but I was basically throwing darts at the wall until it worked.
GiteaMirror commented 2026-04-29 10:16:04 -05:00
Author
Owner
Copy Link

@agentosvg-cmd commented on GitHub (Apr 14, 2026):

I have the same issue. It is so frustrated. Stuck at 77% forever. I have the local model downloaded, but my ollama cant detect the local model downloaded. so i need to run stupid ollama and to download it again.

using macbook air 2024 m4 16gb trying out gemma 4 e4b

<!-- gh-comment-id:4244227700 --> @agentosvg-cmd commented on GitHub (Apr 14, 2026): I have the same issue. It is so frustrated. Stuck at 77% forever. I have the local model downloaded, but my ollama cant detect the local model downloaded. so i need to run stupid ollama and to download it again. using macbook air 2024 m4 16gb trying out gemma 4 e4b
GiteaMirror commented 2026-04-29 10:16:05 -05:00
Author
Owner
Copy Link

@StarGazerDevelopment commented on GitHub (Apr 22, 2026):

@SingularityMan Same as soon as the download starts on a new network, it's fast, but then it slows down really fast.

<!-- gh-comment-id:4300585568 --> @StarGazerDevelopment commented on GitHub (Apr 22, 2026): @SingularityMan Same as soon as the download starts on a new network, it's fast, but then it slows down really fast.
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#56100
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?