Free
$0
For anyone building their first model
- 5 daily-refreshed credits/day (up to 30/mo)
- Models under 5B
- 5 GB model storage
- 250 MB dataset storage
- Manual testing sandbox
Use a model we train on your work, or train your own.
Use a model trained on your work, through one API. It keeps learning from your traffic.
The endpoint is currently invitation only, and we are developing a more streamlined version for wider access.
Per token, by seat or on a dedicated GPU.
$0
For anyone building their first model
$10/mo
For builders who have outgrown Free
$25/mo
For indie devs and solo builders
$50/mo
Most PopularFor power users and client project teams
$100/mo
For scaled teams and multi-client agencies
Custom
For teams training at scale
Prices in USD. Paid plans are billed monthly.
| Ertas Studio | Free | Lite | Builder | Pro | Business |
|---|---|---|---|---|---|
| Monthly price, USD | $0 | $10 | $25 | $50 | $100 |
| Included credits | 5 daily-refreshed credits/day (up to 30/mo) | 40 credits/mo | 100 credits/mo | 200 credits/mo | 400 credits/mo |
| Dataset storage | 250 MB | 1 GB | 2 GB | 5 GB | 10 GB |
| Model storage | 5 GB | 25 GB | 50 GB | 100 GB | 200 GB |
Available on any plan.
$0.30/credit
Only training runs use credits; dataset work and finished-run inference samples are free. Free includes 5 daily-refreshed credits, up to 30 a month. Unused monthly credits on paid plans roll over for 1 to 3 months. You can buy additional credits at $0.30 each on any plan, without a subscription. Bought credits sit outside the monthly usage limit; they last 12 months on Free and never expire on paid plans.
Yes. Upgrades take effect immediately and you’ll be prorated for the remainder of your billing cycle. Downgrades take effect at the next billing cycle.
In Studio, every finished run exports the model files for you to download. On Ertas Endpoint, the model is trained on your work and shaped by your feedback, and it runs on Ertas by default. The model files are handed over to you against a minimum commitment.
Yes. Every completed run exports a GGUF bundle and a LoRA adapter that you download and hand to the client, and they run it on their own infrastructure. Nothing in the artifact is Ertas-branded. If you'd rather not host it, Ertas can serve the model through an API as a managed service, and white-label API access is available case by case.
Yes. Both run on the same Ertas platform, so you can train in Studio and talk to us about running the model on an endpoint when it goes live.