Pricing
Start free. Upgrade when it works.
Every organization starts on Free, with the whole workflow on one model. Plans set how much you build and call. GPUs, storage and model endpoints are billed by their providers, in your own accounts, with no markup from us.
Free
$0forever
Try the whole workflow on one model.
- Training runs
- 1/month
- Model deployments
- 1
- Images (vision)
- Not included
- Members
- 1
- Docs and community support
Start freeNo card needed.
Starter
$49/ month
For individuals shipping a fine-tuned model.
- Training runs
- 5/month
- Model deployments
- 5
- Images (vision)
- Not included
- Members
- 1
- Email support
Start free, upgrade anytimeUpgrade from Settings → Plan and billing.
Pro
Most popular$199/ month
For teams serving models in production.
- Training runs
- Unlimited
- Model deployments
- 1,000
- Images (vision)
- Included
- Members
- 10 included
- Vision model training and serving
- 10 seats included, $30 / month per extra seat
- Priority email support
- Onboarding call
Start free, upgrade anytimeQuestions? Talk to us
Enterprise
Custom
For larger teams, self-hosting or custom limits.
- Training runs
- Custom
- Model deployments
- Custom
- Images (vision)
- Included
- Members
- Custom
- Self-hosted on your own infrastructure
- Custom limits and seats
- Images (vision)
- Dedicated support
Contact usTell us what your team needs.
Compare every limit
Monthly limits count calendar months in UTC and start again on the 1st. A limit a plan does not set is unlimited, except published models and API keys, which stop at 1,000 on every plan. See what each limit counts.
| Limit | Free | Starter | Pro | Enterprise |
|---|---|---|---|---|
| Dataset imports | 1/month | Unlimited | Unlimited | Custom |
| Training runs | 1/month | 5/month | Unlimited | Custom |
| Model deployments | 1 | 5 | 1,000 | Custom |
| API requests | 1K/month | 50K/month | 5M/month | Custom |
| API request rate | 30/minute | 600/minute | 2K/minute | Custom |
| API requests at once | 4 | 16 | 64 | Custom |
| Projects | 1 | 20 | Unlimited | Custom |
| Members | 1 | 1 | 10 included | Custom |
| API keys | 2 | 10 | 1,000 | Custom |
| Images (vision) | Not included | Not included | Included | Included |
Billing questions
Do I need a card to start?
No. Every organization starts on Free, with the whole workflow on one model. A card is asked for only when an owner upgrades.
Does a plan include GPU time?
No. Training runs, deployments and published models run on your own GPU account (RunPod, Verda, Lambda, AWS or Azure), which bills you directly at its own prices. Tensorant adds no markup, and every run and deployment has an hour limit and an hourly price limit you set. See what it costs.
How do I upgrade, switch plans or cancel?
An owner upgrades under Settings → Plan and billing in the console, through a secure checkout with our payment provider, Dodo Payments. Plans renew every month. A switch between Starter and Pro takes effect at once, with a credit for the time left on your current plan. You can cancel at the next billing date or at once; the organization then moves to Free and keeps everything it has. See plans and billing.
When do monthly limits start again?
Imports, training runs and API requests a month count calendar months in UTC and start again on the 1st. An import or a run that fails before it begins never counts.
What happens when I reach a limit?
Nothing you already have is stopped or removed. Adding one more is refused with a message that names the limit, and an owner can upgrade at any time. Past the API request limits, the API answers 429 with a Retry-After header. See when a limit is reached.
How do extra seats work on Pro?
Pro includes 10 seats, and each extra seat is $30 a month. We add them on request: write to [email protected] with the number of seats you need in total.
Do prices include tax?
Yes. Prices include any sales tax or VAT, which our payment provider works out for your country at checkout.
What does Enterprise add?
Custom limits and seats, images, dedicated support, and Tensorant self-hosted on your own infrastructure, at a custom price. Write to [email protected] and tell us what your team needs.
Bring one dataset. Leave with a model you can call.
The Free plan covers the whole workflow on one model, from your first import to a published model.