Activation is required
Installing or running models requires accepted evaluation activation. See Evaluation Keys to activate Primordial for development and device testing.
Downloading and loading are separate
Downloaded means verified model files are installed for offline use. Loaded means the model is in memory and ready to run. Installation never loads a model, and loading never downloads one.
Downloads can continue in the background
On iOS, configure background downloads once as shown in AI Model Setup. Transfers can continue through ordinary suspension and system termination. A user force-quit cancels them.
If the system terminates the app, call the same makeAvailable capability after relaunch.
Primordial reuses completed files and finalizes any remaining model metadata.
Models are shared across clients
Copies of PrimordialClient share one runtime. When they use the same model, Primordial shares
its model weights and coordinates access instead of loading duplicate copies.
Active work protects models
Primordial prevents a model from being unloaded, replaced, or removed while an operation is using it. Loading a model through the advanced API keeps it in memory until you unload it or Primordial responds to system pressure.
Some models cannot run together
Workflows that use your data prepare embeddings before starting generation. Generation and speech models do
not remain in memory together. If active work prevents Primordial from switching models, the operation
returns resourceConflict.
Primordial frees memory automatically
When your app enters the background, the Mac sleeps, or the system reports memory pressure, Primordial unloads models that are not being used. Active work completes or cancels normally and is never removed from memory unexpectedly.
Check whether a model is in memory
let residency = await primordial.advanced.generation.residency()
print(residency.state)Residency is separate from installation: a downloaded model can be unloaded. To update your interface
whenever the model's memory state changes, observe residencyUpdates():
for await residency in primordial.advanced.generation.residencyUpdates() {
updateModelStatus(residency.state)
}The stream immediately provides the model's current state, then reports future changes such as loading and unloading. Stop the observing task when the interface no longer needs updates. Stopping observation does not affect the model.
Show storage used by AI models
Use allocated bytes for the number shown in storage UI. The total counts shared model identities once.
let storage = try await primordial.ai.storageUsage()
print(ByteCountFormatter.string(fromByteCount: storage.total.allocatedBytes, countStyle: .file))For a per-model breakdown, inspect resources or filter by model kind:
for resource in storage.resources(for: .generation) {
print(resource.model.identifier, resource.usage.allocatedBytes)
}Every resource configured for this client is listed, including an inactive generation fallback.
isActive means selected by configuration, not currently loaded in memory. System-owned models
report zero Primordial-managed bytes.