Failed to download or register the following EPs: QNNExecutionProvider

[martinskuta]

I am running the latest foundry version 0.8.117.0 and it doesn't load the QNNExecutionProvider so I cannot use the models on NPU.

There is not much info in the logs apart from:

2026-01-23 17:32:34.290 +01:00 [INF] Provider QNNExecutionProvider download and ensuring ready attempt: Failure
2026-01-23 17:32:34.293 +01:00 [INF] Download attempt for QNNExecutionProvider unsuccessful. Skipping all EP downloads. 
2026-01-23 17:32:34.293 +01:00 [INF] Finished attempt to autoregister certified EPs; finished in 1392ms
2026-01-23 17:32:34.293 +01:00 [INF] Failed to download or register the following EPs: QNNExecutionProvider. Will try installing again later.

My machine is Surface Pro 11, with Snapdragon X1E80100 running Windows 11 Pro Build 26220

When I run foundry model ls I see only models that can run on CPU:

Alias                          Device     Task           File Size    License      Model ID
-----------------------------------------------------------------------------------------------
phi-4                          CPU        chat           10.16 GB     MIT          Phi-4-generic-cpu:1
----------------------------------------------------------------------------------------------------------
phi-3.5-mini                   CPU        chat           2.53 GB      MIT          Phi-3.5-mini-instruct-generic-cpu:1
--------------------------------------------------------------------------------------------------------------------------
phi-3-mini-128k                CPU        chat           2.54 GB      MIT          Phi-3-mini-128k-instruct-generic-cpu:2
-----------------------------------------------------------------------------------------------------------------------------
phi-3-mini-4k                  CPU        chat           2.53 GB      MIT          Phi-3-mini-4k-instruct-generic-cpu:2
---------------------------------------------------------------------------------------------------------------------------
mistral-7b-v0.2                CPU        chat           4.07 GB      apache-2.0   mistralai-Mistral-7B-Instruct-v0-2-generic-cpu:2
---------------------------------------------------------------------------------------------------------------------------------------
deepseek-r1-14b                CPU        chat           11.51 GB     MIT          deepseek-r1-distill-qwen-14b-generic-cpu:3
---------------------------------------------------------------------------------------------------------------------------------
deepseek-r1-7b                 CPU        chat           6.43 GB      MIT          deepseek-r1-distill-qwen-7b-generic-cpu:3
--------------------------------------------------------------------------------------------------------------------------------
qwen2.5-coder-0.5b             CPU        chat, tools    0.80 GB      apache-2.0   qwen2.5-coder-0.5b-instruct-generic-cpu:4
--------------------------------------------------------------------------------------------------------------------------------
phi-4-mini-reasoning           CPU        chat           4.52 GB      MIT          Phi-4-mini-reasoning-generic-cpu:3
-------------------------------------------------------------------------------------------------------------------------
qwen2.5-0.5b                   CPU        chat, tools    0.80 GB      apache-2.0   qwen2.5-0.5b-instruct-generic-cpu:4
--------------------------------------------------------------------------------------------------------------------------
qwen2.5-1.5b                   CPU        chat, tools    1.78 GB      apache-2.0   qwen2.5-1.5b-instruct-generic-cpu:4
--------------------------------------------------------------------------------------------------------------------------
qwen2.5-coder-1.5b             CPU        chat, tools    1.78 GB      apache-2.0   qwen2.5-coder-1.5b-instruct-generic-cpu:4
--------------------------------------------------------------------------------------------------------------------------------
phi-4-mini                     CPU        chat, tools    4.80 GB      MIT          Phi-4-mini-instruct-generic-cpu:5
------------------------------------------------------------------------------------------------------------------------
qwen2.5-14b                    CPU        chat, tools    11.06 GB     apache-2.0   qwen2.5-14b-instruct-generic-cpu:4
-------------------------------------------------------------------------------------------------------------------------
qwen2.5-coder-14b              CPU        chat, tools    11.06 GB     apache-2.0   qwen2.5-coder-14b-instruct-generic-cpu:4
-------------------------------------------------------------------------------------------------------------------------------
qwen2.5-coder-7b               CPU        chat, tools    6.16 GB      apache-2.0   qwen2.5-coder-7b-instruct-generic-cpu:4
------------------------------------------------------------------------------------------------------------------------------
qwen2.5-7b                     CPU        chat, tools    6.16 GB      apache-2.0   qwen2.5-7b-instruct-generic-cpu:4
------------------------------------------------------------------------------------------------------------------------
gpt-oss-20b                    CPU        chat           12.26 GB     MIT          gpt-oss-20b-generic-cpu:1

When I install older version of the foundry winget install --id=Microsoft.FoundryLocal -v "0.6.87.59034" -e it seems to work:

Alias                          Device     Task               File Size    License      Model ID
-----------------------------------------------------------------------------------------------
phi-4                          CPU        chat-completion    10.16 GB     MIT          Phi-4-generic-cpu
--------------------------------------------------------------------------------------------------------
phi-3.5-mini                   NPU        chat-completion    2.78 GB      MIT          phi-3.5-mini-instruct-qnn-npu
                               CPU        chat-completion    2.53 GB      MIT          Phi-3.5-mini-instruct-generic-cpu
------------------------------------------------------------------------------------------------------------------------
deepseek-r1-14b                NPU        chat-completion    7.12 GB      MIT          deepseek-r1-distill-qwen-14b-qnn-npu
---------------------------------------------------------------------------------------------------------------------------
deepseek-r1-7b                 NPU        chat-completion    3.71 GB      MIT          deepseek-r1-distill-qwen-7b-qnn-npu
--------------------------------------------------------------------------------------------------------------------------
phi-3-mini-128k                NPU        chat-completion    2.78 GB      MIT          phi-3-mini-128k-instruct-qnn-npu
                               CPU        chat-completion    2.54 GB      MIT          Phi-3-mini-128k-instruct-generic-cpu
---------------------------------------------------------------------------------------------------------------------------
phi-3-mini-4k                  NPU        chat-completion    2.78 GB      MIT          phi-3-mini-4k-instruct-qnn-npu
                               CPU        chat-completion    2.53 GB      MIT          Phi-3-mini-4k-instruct-generic-cpu
-------------------------------------------------------------------------------------------------------------------------
mistral-7b-v0.2                CPU        chat-completion    4.07 GB      apache-2.0   mistralai-Mistral-7B-Instruct-v0-2-generic-cpu
-------------------------------------------------------------------------------------------------------------------------------------
whisper-base                   CPU        automatic-speech-recognition 0.37 GB      apache-2.0   openai-whisper-base-generic-cpu
----------------------------------------------------------------------------------------------------------------------
whisper-medium                 CPU        automatic-speech-recognition 3.05 GB      apache-2.0   openai-whisper-medium-generic-cpu
------------------------------------------------------------------------------------------------------------------------
whisper-small                  CPU        automatic-speech-recognition 1.05 GB      apache-2.0   openai-whisper-small-generic-cpu
-----------------------------------------------------------------------------------------------------------------------
whisper-tiny                   CPU        automatic-speech-recognition 0.22 GB      apache-2.0   openai-whisper-tiny-generic-cpu
----------------------------------------------------------------------------------------------------------------------
qwen2.5-coder-0.5b             CPU        chat-completion    0.80 GB      apache-2.0   qwen2.5-coder-0.5b-instruct-generic-cpu
------------------------------------------------------------------------------------------------------------------------------
phi-4-mini-reasoning           CPU        chat-completion    4.52 GB      MIT          Phi-4-mini-reasoning-generic-cpu
-----------------------------------------------------------------------------------------------------------------------
qwen2.5-0.5b                   CPU        chat-completion    0.80 GB      apache-2.0   qwen2.5-0.5b-instruct-generic-cpu
------------------------------------------------------------------------------------------------------------------------
qwen2.5-1.5b                   NPU        chat-completion    2.78 GB      MIT          qwen2.5-1.5b-instruct-qnn-npu
                               CPU        chat-completion    1.78 GB      apache-2.0   qwen2.5-1.5b-instruct-generic-cpu
------------------------------------------------------------------------------------------------------------------------
qwen2.5-coder-1.5b             CPU        chat-completion    1.78 GB      apache-2.0   qwen2.5-coder-1.5b-instruct-generic-cpu
------------------------------------------------------------------------------------------------------------------------------
qwen2.5-14b                    CPU        chat-completion    11.06 GB     apache-2.0   qwen2.5-14b-instruct-generic-cpu
-----------------------------------------------------------------------------------------------------------------------
qwen2.5-coder-14b              CPU        chat-completion    11.06 GB     apache-2.0   qwen2.5-coder-14b-instruct-generic-cpu
-----------------------------------------------------------------------------------------------------------------------------
qwen2.5-coder-7b               CPU        chat-completion    6.16 GB      apache-2.0   qwen2.5-coder-7b-instruct-generic-cpu
----------------------------------------------------------------------------------------------------------------------------
qwen2.5-7b                     NPU        chat-completion    2.78 GB      MIT          qwen2.5-7b-instruct-qnn-npu
                               CPU        chat-completion    6.16 GB      apache-2.0   qwen2.5-7b-instruct-generic-cpu
----------------------------------------------------------------------------------------------------------------------
whisper-large-v3-turbo         CPU        automatic-speech-recognition 8.79 GB      apache-2.0   openai-whisper-large-v3-turbo-generic-cpu
--------------------------------------------------------------------------------------------------------------------------------
gpt-oss-20b                    CPU        chat-completion    12.26 GB     MIT          gpt-oss-20b-generic-cpu

I have a tingle that it is related to the fact that I have windows workload execution provider that comes with windows, because if I run Get-AppxPackage -AllUsers "*.EP.*" | Select-Object -ExpandProperty PackageFullName I get this:

Windows.Workload.EP.Qualcomm.QNN.1.8_1.8.27.0_arm64__8wekyb3d8bbwe
WindowsWorkload.EP.Qualcomm.QNN.1.8_1.8.31.0_arm64__8wekyb3d8bbwe

but there is no way to remove those, because running Get-AppxPackage -AllUsers "WindowsWorkload.EP*" | Remove-AppxPackage as admin or as AUTHORITY/SYSTEM seems to start something in the terminal, but it disappears immediatelly without any result.

Any help would be appreciated. Thx.

[natke]

Hi @martinskuta,

Can you please try the Get-AppxPackage -AllUsers "*.EP.*" | Select-Object -ExpandProperty PackageFullName again after you have run the Get-AppxPackage -AllUsers "WindowsWorkload.EP*" | Remove-AppxPackage command? You might also need an -AllUsers in your Remove-AppxPackage command.

Then try foundry service init.

If the EP still fails to download and register, can you please try the following:

remove-appxpackage -AllUsers uup://product/Windows.Workload.ExecutionProvider.QNN.arm64

Then try foundry service init again.

[martinskuta]

Hi @natke , thx for reply. I tried it, but it doesn't remove it without any error or any apparent indication of what happened, it just returns immediately. I feel like the Windows.Workload* packages are impossible to remove or I didn't find yet a way :) Here is the output from Terminal:

>> Get-AppxPackage -AllUsers "*.EP.*"  | Select-Object -ExpandProperty PackageFullName
Windows.Workload.EP.Qualcomm.QNN.1.8_1.8.27.0_arm64__8wekyb3d8bbwe
WindowsWorkload.EP.Qualcomm.QNN.1.8_1.8.31.0_arm64__8wekyb3d8bbwe
>> Get-AppxPackage -AllUsers "WindowsWorkload.EP*" | Remove-AppxPackage -AllUsers
>> Get-AppxPackage -AllUsers "*.EP.*"  | Select-Object -ExpandProperty PackageFullName
Windows.Workload.EP.Qualcomm.QNN.1.8_1.8.27.0_arm64__8wekyb3d8bbwe
WindowsWorkload.EP.Qualcomm.QNN.1.8_1.8.31.0_arm64__8wekyb3d8bbwe
>> foundry service init
🟢 Service is Started on http://127.0.0.1:50195/, PID 4000!
🕚 Downloading complete!...
Failed to download or register the following EPs: QNNExecutionProvider. Will try installing again later.
Valid EPs: CPUExecutionProvider
>> remove-appxpackage -AllUsers uup://product/Windows.Workload.ExecutionProvider.QNN.arm64
>> foundry service init
🕚 Downloading complete!...
Failed to download or register the following EPs: QNNExecutionProvider. Will try installing again later.
Valid EPs: CPUExecutionProvider
>>

it is also weird that I have two packages, which appear to be the same, for the EP, with only a . missing in the WindowsWorkload part of the name, no idea how that happened, because I have installed windows fresh not so long ago.