This documentation is for the upcoming version of the AI Manager
Read the documentation for the current version
Ctrlk
AI ManagerNx Meta
For the complete documentation index, see llms.txt. This page is also available as Markdown.

Upload your model

Step-by-step guide to uploading a model file through the Add a model wizard.

After creating your model, upload it to AI Manager Cloud to make it available on all your Nx Servers. Click the + icon in the sidebar to open the Add a model wizard.

The wizard walks you through five steps.

Step 1: Upload a Model File

Drag and drop your model file into the upload area, or click the area to select a file from your file system. Supported formats include ONNX, TFLite, and ZIP (Teachable Machine, NetsPresso).

Add a model wizard showing the Upload a Model File step with a drag-and-drop area and supported format icons below
Step 1: drag and drop your model file or click to browse. Supported formats are listed below the upload area.

Once a file is selected, the filename appears in the upload area. Click Next to continue.

Add a model wizard with a model file selected, showing the filename and a Click to change file link
File selected: click Next to proceed to owner selection.

Step 2: Set an Owner

If your account belongs to multiple organizations or channel partners, choose which one will own the model. If your account has only one organization, it is pre-selected; click Next to continue.

Add a model wizard showing a list of channel partners and organizations to choose from, each with a Select button
Select the owner organization or channel partner for this model.
Add a model wizard with a single organization pre-selected and a Change Association button
If only one organization is available, it is selected automatically.

Step 3: Catalogue

Optionally assign the model to a catalogue so it is grouped with related models and visible to specific sites or organizations. Type to search for an existing catalogue, or leave the field empty to skip.

Add a model wizard showing the Catalogue step with an empty search field and a note that new catalogues can be created from the models page
Search for a catalogue or leave blank to skip. New catalogues can be created from the Models page.
Add a model wizard with a catalogue selected in the dropdown field
Catalogue selected: the model will appear in this catalogue after upload.

Click Next to continue.

Step 4: Hardware Support

Choose which additional hardware conversions to enable. Each enabled conversion generates an optimized model artifact for a specific AI accelerator and consumes compute tokens from your account.

Add a model wizard showing the Additional hardware support step with Default ONNX checked and a note that enabled conversions will require compute tokens
Select the hardware targets you need. Default ONNX is always enabled and covers CPU, Intel, and NVIDIA hardware.

If your model needs to run on Hailo or other specialized accelerators, enable those conversions here. Click Next to continue.

Step 5: Model Details

Give your model a name and add documentation describing what it does, its input requirements, and any relevant notes. Both fields are required. The name is pre-filled from the filename; edit it to something descriptive.

Add a model wizard showing the Model details step with Name and Documentation fields filled in, and an Upload button at the bottom
Enter a clear name and documentation for your model, then click Upload.

Click Upload to start the transfer.

Uploading and Processing

A progress bar shows the upload status. After the file is transferred, the cloud platform converts it for each selected hardware target. This can take several minutes depending on the model size and the number of enabled conversions.

Upload progress dialog showing a progress bar at 67% with the message Your model is being uploaded and will be processed shortly
The model is uploading: keep the page open until the progress reaches 100%.

When the upload completes you are taken to the model detail page. A notification confirms the upload and indicates that an email is sent when conversion finishes.

Model detail page showing a Model upload completed notification and the model status, catalogues, and owner information
Model upload complete: conversion runs in the background, and an email arrives when the model is ready to deploy.

Once conversion is complete and the model status shows ok, the model is ready to be used on your devices.

If your model uses normalization values that differ from the defaults, set them from the model detail page before deploying.

PreviousCloud Quick StartNextNormalization

Last updated