github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-06 16:11:34 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #14279] Qwen3.5-397B-A17B Cloud data retention and privacy concerns #9296

New Issue
Open
opened 2026-04-12 22:09:48 -05:00 by GiteaMirror · 4 comments
GiteaMirror commented 2026-04-12 22:09:48 -05:00
Owner
Copy Link

Originally created by @asitwere on GitHub (Feb 16, 2026).
Original GitHub issue: https://github.com/ollama/ollama/issues/14279

It looks like Alibaba is currently the only endpoint available for Qwen3.5, but Ollama's docs/advertising for Ollama Cloud provide data privacy assurances.

Since Alibaba retains prompts & responses, can it be confirmed that users are not being routed to Alibaba APIs via Ollama Cloud?

Originally created by @asitwere on GitHub (Feb 16, 2026). Original GitHub issue: https://github.com/ollama/ollama/issues/14279 It looks like Alibaba is currently the only endpoint available for Qwen3.5, but Ollama's docs/advertising for Ollama Cloud provide data privacy assurances. Since Alibaba retains prompts & responses, can it be confirmed that users are not being routed to Alibaba APIs via Ollama Cloud?
GiteaMirror added the question label 2026-04-12 22:09:48 -05:00
GiteaMirror commented 2026-04-12 22:09:49 -05:00
Author
Owner
Copy Link

@teddybear082 commented on GitHub (Feb 17, 2026):

What are you talking about? Qwen 3.5 model is on hugging face.

<!-- gh-comment-id:3913803756 --> @teddybear082 commented on GitHub (Feb 17, 2026): What are you talking about? Qwen 3.5 model is on hugging face.
GiteaMirror commented 2026-04-12 22:09:49 -05:00
Author
Owner
Copy Link

@asitwere commented on GitHub (Feb 17, 2026):

What are you talking about? Qwen 3.5 model is on hugging face.

Yes, you can pull Qwen 3.5 from HuggingFace and run it locally.

The problem is that Ollama is calling third-party APIs for their Cloud models. There is nothing in the docs claiming that Ollama controls the infrastructure for their Cloud models. There is no ZDR guarantee. Alibaba Cloud is still the only API endpoint for Qwen 3.5 on OpenRouter.

Ollama is unclear in their docs, which only say "Keep your data private" and "Ollama does not record, log or train on any prompt or response data." There is nothing clarifying that Ollama controls the end-to-end infrastructure. If they don't actually host the model and serve the APIs themselves (for Cloud), these statements are incomplete/misleading.

If this wasn't an issue it would be clear in the docs.

<!-- gh-comment-id:3914270617 --> @asitwere commented on GitHub (Feb 17, 2026): > What are you talking about? Qwen 3.5 model is on hugging face. Yes, you can pull Qwen 3.5 from HuggingFace and run it locally. The problem is that Ollama is calling third-party APIs for their Cloud models. There is nothing in the docs claiming that Ollama controls the infrastructure for their Cloud models. There is no ZDR guarantee. Alibaba Cloud is still the only API endpoint for Qwen 3.5 on OpenRouter. Ollama is unclear in their docs, which only say "Keep your data private" and "Ollama does not record, log or train on any prompt or response data." There is nothing clarifying that Ollama controls the end-to-end infrastructure. If they don't actually host the model and serve the APIs themselves (for Cloud), these statements are incomplete/misleading. If this wasn't an issue it would be clear in the docs.
GiteaMirror commented 2026-04-12 22:09:50 -05:00
Author
Owner
Copy Link

@omid3098 commented on GitHub (Apr 10, 2026):

I also would like to know about this. I am about to set up a personal VPS with Open WebUI and use my Ollama Pro account for it, and I want to know where my API calls will be processed. encrypted on Ollama's own servers running on NVIDIA hardware, or routed elsewhere.
According to the FAQ, Ollama states: "We don't store your prompts or outputs" and "Ollama does not record, log or train on any prompt or response data." The privacy policy also mentions that cloud-hosted content is processed "transiently" and not retained beyond the time required to fulfill the request.
However, none of this clarifies whether certain cloud models are served directly by the original model providers (e.g., Alibaba for Qwen, Zhipu for GLM). If that's the case, their data handling policies would apply, not Ollama's.

Could the team confirm:

  1. Are all cloud model requests processed exclusively on Ollama-managed infrastructure?
  2. If third-party inference providers are involved, which ones, and do Ollama's privacy guarantees still apply end-to-end?
<!-- gh-comment-id:4227214358 --> @omid3098 commented on GitHub (Apr 10, 2026): I also would like to know about this. I am about to set up a personal VPS with Open WebUI and use my Ollama Pro account for it, and I want to know where my API calls will be processed. encrypted on Ollama's own servers running on NVIDIA hardware, or routed elsewhere. According to the FAQ, Ollama states: "We don't store your prompts or outputs" and "Ollama does not record, log or train on any prompt or response data." The privacy policy also mentions that cloud-hosted content is processed "transiently" and not retained beyond the time required to fulfill the request. However, none of this clarifies whether certain cloud models are served directly by the original model providers (e.g., Alibaba for Qwen, Zhipu for GLM). If that's the case, their data handling policies would apply, not Ollama's. Could the team confirm: 1. Are all cloud model requests processed exclusively on Ollama-managed infrastructure? 2. If third-party inference providers are involved, which ones, and do Ollama's privacy guarantees still apply end-to-end?
GiteaMirror commented 2026-04-12 22:09:50 -05:00
Author
Owner
Copy Link

@asitwere commented on GitHub (Apr 12, 2026):

There have been a few updates to the FAQ on the Pricing page:

PRIVACY

Where are models hosted?

Ollama hosts models and compute resources primarily in the United States. To serve global demand, we may route to Europe and Singapore for additional capacity.

Is my prompt or response data trained on?

Prompt or response data is never logged or trained on

Who does Ollama partner with to host models?

Ollama collaborates with NVIDIA Cloud Providers (NCPs) to host open models.
When Ollama partners with providers, we require no logging, no training, and zero data retention policies in place.

SOURCE: https://ollama.com/pricing

<!-- gh-comment-id:4231582569 --> @asitwere commented on GitHub (Apr 12, 2026): There have been a few updates to the FAQ on the Pricing page: PRIVACY Where are models hosted? Ollama hosts models and compute resources primarily in the United States. To serve global demand, we may route to Europe and Singapore for additional capacity. Is my prompt or response data trained on? Prompt or response data is never logged or trained on Who does Ollama partner with to host models? Ollama collaborates with NVIDIA Cloud Providers (NCPs) to host open models. When Ollama partners with providers, we require no logging, no training, and zero data retention policies in place. SOURCE: https://ollama.com/pricing
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 question
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#9296
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?