[GH-ISSUE #14535] qwen3.5:27b will not work with Android Studio because of metadata issue #35190

Open
opened 2026-04-22 19:33:16 -05:00 by GiteaMirror · 3 comments
Owner

Originally created by @chedabob on GitHub (Mar 1, 2026).
Original GitHub issue: https://github.com/ollama/ollama/issues/14535

What is the issue?

When trying to connect to Ollama with the new Android Studio AI support for local models, I get the following error when it tries to query Ollama's API for available models:

java.lang.IllegalStateException: Expected a string but was BEGIN_ARRAY at line 1 column 24906 path $.model_info.
See https://github.com/google/gson/blob/main/Troubleshooting.md#unexpected-json-structure

I believe it's using http://localhost:11434/api/show to get the model details, and in the returned response the following line is what is causing the issue

"qwen35.vision.deepstack_visual_indexes": []

I'm not sure if this an issue with the Ollama metadata, the API itself, or a bug in the IntelliJ code not handling something that's part of the schema.

This seems to be the case for both qwen3.5:27b and qwen3.5:35b

Relevant log output


OS

macOS

GPU

Apple

CPU

Apple

Ollama version

Version 0.17.4 (0.17.4)

Originally created by @chedabob on GitHub (Mar 1, 2026). Original GitHub issue: https://github.com/ollama/ollama/issues/14535 ### What is the issue? When trying to connect to Ollama with the new Android Studio AI support for local models, I get the following error when it tries to query Ollama's API for available models: > java.lang.IllegalStateException: Expected a string but was BEGIN_ARRAY at line 1 column 24906 path $.model_info. See https://github.com/google/gson/blob/main/Troubleshooting.md#unexpected-json-structure I believe it's using `http://localhost:11434/api/show` to get the model details, and in the returned response the following line is what is causing the issue `"qwen35.vision.deepstack_visual_indexes": []` I'm not sure if this an issue with the Ollama metadata, the API itself, or a bug in the IntelliJ code not handling something that's part of the schema. This seems to be the case for both `qwen3.5:27b` and `qwen3.5:35b` ### Relevant log output ```shell ``` ### OS macOS ### GPU Apple ### CPU Apple ### Ollama version Version 0.17.4 (0.17.4)
GiteaMirror added the bug label 2026-04-22 19:33:16 -05:00
Author
Owner

@sarn0ld commented on GitHub (Mar 4, 2026):

Removing the qwen3.5 models fixes the list in Android Studio.
qwen3-vl is working: https://ollama.com/library/qwen3-vl

<!-- gh-comment-id:3996000966 --> @sarn0ld commented on GitHub (Mar 4, 2026): Removing the qwen3.5 models fixes the list in Android Studio. qwen3-vl is working: https://ollama.com/library/qwen3-vl
Author
Owner

@tsnrjohnson-cloud commented on GitHub (Mar 30, 2026):

I'm also having this issue, and it's impactful.
We should not need to remove one of the best-performing models entirely from the list to have it load correctly...

<!-- gh-comment-id:4155774617 --> @tsnrjohnson-cloud commented on GitHub (Mar 30, 2026): I'm also having this issue, and it's impactful. We should not need to remove one of the best-performing models entirely from the list to have it load correctly...
Author
Owner

@svenjacobs commented on GitHub (Apr 3, 2026):

Someone reported this to Google and it has already been fixed in Android Studio. Unfortunately, the fix is not yet available in the current stable version Panda 3 (2025.3.3) so we have to wait a bit or use the canary version.

<!-- gh-comment-id:4182979919 --> @svenjacobs commented on GitHub (Apr 3, 2026): Someone [reported this to Google](https://issuetracker.google.com/issues/490118031) and it has already been fixed in Android Studio. Unfortunately, the fix is not yet available in the current stable version Panda 3 (2025.3.3) so we have to wait a bit or use the canary version.
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#35190