Mintplex-Labs/anything-llm
1
0
Fork 0
mirror of https://github.com/Mintplex-Labs/anything-llm.git synced 2026-07-19 14:13:35 -04:00
Code Issues 328 Packages Projects Releases 10 Wiki Activity

[GH-ISSUE #4906] [BUG]: Meeting Assistant GPU mode fails - Tinyscribe cannot start (CUDA runtime not bundled) #3085

New Issue
Closed
opened 2026-02-22 18:32:33 -05:00 by yindo · 1 comment
yindo commented 2026-02-22 18:32:33 -05:00
Owner
Copy Link
Copy Source

Originally created by @xxx-xxx-xxx-xxx-xxx-xxx on GitHub (Jan 27, 2026).
Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/4906

How are you running AnythingLLM?

AnythingLLM desktop app

What happened?

When I select GPU mode in Meeting Assistant Preferences, it downloads the runtime libraries and shows "Downloaded". However, when I try to process audio, Tinyscribe fails to start with "ECONNREFUSED" error because it cannot find CUDA runtime libraries on my system.

System Information:

  • Windows 10
  • GPU: RTX 5060 Ti with latest drivers (591.74)
  • AnythingLLM version: 1.10.0
  • Other GPU tools (LM Studio, Jan.ai) work fine

Error from logs:

[MeetingAssistantHelper] Expecting Tinyscribe server executable to be at: C:\Users<username>\AppData\Roaming\anythingllm-desktop\storage\engines\meeting-assistant\windows-x64-cuda\tinyscribe.exe

[MeetingAssistantHelper] master-recording.wav - Transcription failed! TypeError: fetch failed
[cause]: AggregateError [ECONNREFUSED]

Expected behavior:
Either:

  1. AnythingLLM should bundle CUDA runtime libraries with the GPU download, OR
  2. Show a clear error message: "Please install CUDA Toolkit 12.x and cuDNN before using GPU mode"

Current workaround:
GPU mode only works after manually installing CUDA Toolkit and cuDNN system-wide.

Impact:
This is confusing for users because the "Downloaded" status implies GPU is ready to use, but it silently fails without explanation.

Are there known steps to reproduce?

No response

Originally created by @xxx-xxx-xxx-xxx-xxx-xxx on GitHub (Jan 27, 2026). Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/4906 ### How are you running AnythingLLM? AnythingLLM desktop app ### What happened? When I select GPU mode in Meeting Assistant Preferences, it downloads the runtime libraries and shows "Downloaded". However, when I try to process audio, Tinyscribe fails to start with "ECONNREFUSED" error because it cannot find CUDA runtime libraries on my system. **System Information:** - Windows 10 - GPU: RTX 5060 Ti with latest drivers (591.74) - AnythingLLM version: 1.10.0 - Other GPU tools (LM Studio, Jan.ai) work fine **Error from logs:** [MeetingAssistantHelper] Expecting Tinyscribe server executable to be at: C:\Users\<username>\AppData\Roaming\anythingllm-desktop\storage\engines\meeting-assistant\windows-x64-cuda\tinyscribe.exe [MeetingAssistantHelper] master-recording.wav - Transcription failed! TypeError: fetch failed [cause]: AggregateError [ECONNREFUSED] **Expected behavior:** Either: 1. AnythingLLM should bundle CUDA runtime libraries with the GPU download, OR 2. Show a clear error message: "Please install CUDA Toolkit 12.x and cuDNN before using GPU mode" **Current workaround:** GPU mode only works after manually installing CUDA Toolkit and cuDNN system-wide. **Impact:** This is confusing for users because the "Downloaded" status implies GPU is ready to use, but it silently fails without explanation. ### Are there known steps to reproduce? _No response_
yindo added the possible bug label 2026-02-22 18:32:33 -05:00
yindo closed this issue 2026-02-22 18:32:33 -05:00
yindo commented 2026-02-22 18:32:33 -05:00
Author
Owner
Copy Link
Copy Source

@timothycarambat commented on GitHub (Jan 27, 2026):

AnythingLLM should bundle CUDA runtime libraries with the GPU download, OR

We do, we bundle the torch required bins for CUDA compatibility, but you say this only resolved once you installed the CUDA toolkit? The toolkit is several GB so we will have to look into this, moving to internal board for this 👍

@timothycarambat commented on GitHub (Jan 27, 2026): > AnythingLLM should bundle CUDA runtime libraries with the GPU download, OR We do, we bundle the torch required bins for CUDA compatibility, but you say this only resolved once you installed the CUDA toolkit? The toolkit is several GB so we will have to look into this, moving to internal board for this 👍
yindo changed title from [BUG]: Meeting Assistant GPU mode fails - Tinyscribe cannot start (CUDA runtime not bundled) to [GH-ISSUE #4906] [BUG]: Meeting Assistant GPU mode fails - Tinyscribe cannot start (CUDA runtime not bundled) 2026-06-05 14:50:13 -04:00
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
Desktop
Docker
Integration Request
Integration Request
OS: Linux
OS: Mobile
OS: Windows
UI/UX
blocked
bug
bug
core-team-only
documentation
duplicate
embed-widget
enhancement
feature request
github_actions
good first issue
investigating
needs info / can't replicate
possible bug
pull-request
Mirrored from GitHub Pull Request
 question
stage: specifications
wontfix
No Label possible bug
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
yindo
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: Mintplex-Labs/anything-llm#3085
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?