Simple, Scalable, Transparent
All plans give you access to the TAUS Generic Model.
You can choose to use QE only or QE + APE.
Up to 500,000 QE characters
EN, FR, IT, DE, ES
valid for 1 month
€
0
Unlimited characters
100+ languages
Monthly pay-per-use billing
€
500
Up to 3.125M QE characters
100+ languages
valid for 12 months
€
2,000
Up to 12.5M QE characters
100+ languages
valid for 12 months
€
5,000
Up to 31.25M QE characters
100+ languages
valid for 12 months
Access to the TAUS Generic Model
Use what you need: QE alone or QE + APE
QE: €0.0002 per source and target character
APE: €0.01 per source word
If you have an existing prepaid plan, it will be used first for any QE usage. To switch to a usage-based plan, simply deplete your current prepaid credits.
Contact us for an offer tailored to your needs.
How do we charge QE and APE?
QE and APE can be used in either a real-time or batch mode (asynchronous). Unit price varies depending on which mode is used.
Use this API when your use case requires instant results or user-facing interactions.
€
0,0002
per source and target character (QE)
€
0,01
per source word (APE)
Use this API for high-volume, non-interactive workflows where instant responses are not required.
€
0.00016
per source and target character (QE)
€
0.008
per source word (APE)
For true flexibility, opt for a usage-based plan which offers both QE and APE. Billing is simple and predictable, calculated at the end of each month based on actual usage.
€
0.0002
€
0.01
Click here to obtain an API key for usage-based billing.
Feel free to reach out if you need assistance selecting the best plan for your needs.
Important Note: If you have an existing prepaid plan, it will be used first for any QE usage. To switch to a usage-based plan, simply deplete your current prepaid credits.
I’m not sure which plan fits my needs best
Are there any upfront costs?
Can we track our EPIC API usage?
What happens if I run out of credits?
Is technical support available?
I'm interested in a custom model, how does that work?
What kind of data do you need to train a custom model?