Free
$0 per month
30 GPU-minutes a month
Enough GPU time for the full first round trip and a real first project.
Every plan is the same product. Plans differ in one thing: how many GPU-minutes of training you get, which is the time our GPUs spend building your models.
Prices are provisional, and nothing is charged yet.
Billing is in test mode while we measure what training really costs us. The prices below are our current estimates and may change before billing opens.
$0 per month
30 GPU-minutes a month
Enough GPU time for the full first round trip and a real first project.
Our pick for one person
$19 per month
300 GPU-minutes a month
For one person building several projects a month.
$99 per month
1,500 GPU-minutes a month
For a small group sharing one pool of GPU time.
$399 per month
6,000 GPU-minutes a month
For a company that needs a signed Data Processing Addendum.
$299 per year
3,000 GPU-minutes a year, for the school year
For one classroom, for the school year.
One minute of our GPUs training a model for you. Each training run is timed from start to finish and rounded up to the next whole minute: a 40-second run counts as 1 minute, a 61-second run as 2.
New training runs are declined, with a message asking you to upgrade your plan. A run that has already started finishes.
Everything else keeps working: you can still collect and label data, review your results and download code for the models you have trained. A paid plan's allowance starts again at the beginning of each billing period.
Time our GPUs spend training a model for you. Each training run is timed from start to finish and rounded up to the next whole minute. Collecting, labelling and downloading code for a board use none.
Yes, at any time, by writing to support@intelligena.com. Cancelling stops future charges and takes effect at the end of the period you have paid for; that period is not refunded. Your projects, recordings and models are not deleted when a plan ends.
No. We do not use your recordings to train, fine-tune or improve any third-party general-purpose or foundation model, and our contracts forbid our providers from doing so. We may use aggregated data that cannot identify you to improve our own systems. Under the Data Processing Addendum, a model trained for you on your own data is your content.
In the United States and its territories. That includes every backup, replica and cache, including copies held by our hosting and storage providers.
Not to collect data or train: both happen in your browser, and your phone records without an app. To put a model on your board you build the downloaded C code with that board's usual tools, such as the Arduino IDE for the Nano 33 BLE and the ESP32-S3.
No card needed. Check that your board is supported, then train your first model.