Open in the explorer

DictationModel

Type
Computer and user
GPO path
Computer Configuration › Policies › Administrative Templates › Visual Studio Code › Chat User Configuration › Policies › Administrative Templates › Visual Studio Code › Chat
Supported on
Visual Studio Code >= 1.136
ADMX file
VSCode.admx
Internal name
VSCode/DictationModel

Explanation

Controls the transcription model used for dictation.

Registry

HKEY_LOCAL_MACHINE\Software\Policies\Microsoft\VSCode
HKEY_CURRENT_USER\Software\Policies\Microsoft\VSCode
Value name
not defined in the ADMX

Enable / disable values: not defined in the ADMX (see the options below).

PowerShell

PowerShell commands that write to the registry what the policy would write. Values in angle brackets (<…>) must be replaced.

Writing directly to the registry bypasses Group Policy: if a GPO manages this value, it will be rewritten at the next gpupdate; otherwise the value stays in place (the "tattooing" effect). Prefer a GPO or, locally, LGPO.exe. Run PowerShell as administrator for HKLM. Some policies need a restart or gpupdate /force.

Computer (HKLM)

Enable

# Enable: VSCode/DictationModel
$k = 'HKLM:\Software\Policies\Microsoft\VSCode'
if (-not (Test-Path -Path $k)) { New-Item -Path $k -Force | Out-Null }
#
# Possible options:
#   Use the on-device transcription model. Audio does not leave the device. = nemotron-3.5-asr-streaming-0.6b
#   Use the cloud transcription service. Audio is streamed to the service. = mai
Set-ItemProperty -Path $k -Name 'DictationModel' -Value 'nemotron-3.5-asr-streaming-0.6b' -Type String

Disable

# Disable: VSCode/DictationModel
# No value is written in the "Disabled" state (nothing defined in the ADMX)

Revert to "Not configured"

# Revert to "Not configured": VSCode/DictationModel
$k = 'HKLM:\Software\Policies\Microsoft\VSCode'
Remove-ItemProperty -Path $k -Name 'DictationModel' -ErrorAction SilentlyContinue

User (HKCU)

Enable

Disable

Revert to "Not configured"

Options (1)

LabelControlRegistry valueREG typeConstraints / details
DictationModelDrop-down listenumDictationModelREG_SZ
LabelValue
Use the on-device transcription model. Audio does not leave the device.REG_SZ = nemotron-3.5-asr-streaming-0.6b
Use the cloud transcription service. Audio is streamed to the service.REG_SZ = mai

Application

Visual Studio Code (Microsoft) — November 2026 templates

Official templates downloaded on November 12, 2026.