The Models workspace is the main place to manage local model files and runtime readiness.

​
Web UI

Open Models from the bottom of the sidebar. The page shows:
  • Total visible models
  • Downloaded and loaded counts
  • Total local model storage used
  • Provider-grouped model cards
  • Search, status filters, and category filters
Status filters:
FilterMeaning
AllShow every enabled catalog model.
LoadedShow models with ready status.
DownloadedShow models downloaded to disk but not loaded.
Not downloadedShow models available for download.
Category filters:
  • Text to Speech
  • Transcription
  • Chat

​
Model Statuses

StatusMeaningTypical action
not_downloadedAvailable in the catalog, not on diskDownload
downloadingDownload in progressWait or cancel
downloadedFiles are on disk, model is not in memoryLoad or delete
loadingRuntime is loading the modelWait
readyModel is loaded and available for inferenceUse or unload
errorDownload/load failedRefresh, retry, or inspect logs

​
Actions

ActionWhereNotes
RefreshModels page headerRe-reads catalog and local status.
DownloadModel cardDownloads model files into the configured models directory.
CancelDownloading model cardCancels an active download when the runtime supports cancellation.
LoadDownloaded model cardLoads the model into memory.
UnloadReady model cardFrees runtime memory while keeping files on disk.
DeleteDownloaded or ready model cardRemoves local model files after confirmation.
Gemma models may show Manual DL because they require manual Hugging Face access setup. See Manual Model Downloads.

​
CLI

# List catalog models
izwi list

# List downloaded models only
izwi list --local

# Download
izwi pull Qwen3-TTS-12Hz-0.6B-Base

# Load and unload
izwi models load Qwen3-TTS-12Hz-0.6B-Base
izwi models unload Qwen3-TTS-12Hz-0.6B-Base

# Delete files
izwi rm Qwen3-TTS-12Hz-0.6B-Base

# Runtime status
izwi status --detailed

​
API

Model management routes live under /v1/admin/models:
RoutePurpose
GET /v1/admin/modelsList model catalog and local status.
POST /v1/admin/models/{variant}/downloadStart a download.
GET /v1/admin/models/{variant}/download/progressRead download progress.
POST /v1/admin/models/{variant}/download/cancelCancel a download.
POST /v1/admin/models/{variant}/loadLoad into memory.
POST /v1/admin/models/{variant}/unloadUnload from memory.
DELETE /v1/admin/models/{variant}Delete local files.
See the API Reference for response shapes and capability fields.

​
See Also