We build AI products that make intelligence affordable.
TalkingMachines is an Estonian software company developing eterial.ai — a platform for low-cost AI inference.
eterial.ai — one platform, two sides
An AI inference platform for developers and everyone else, built on the same low-cost engine.
Inference API
An OpenAI-compatible API for production AI workloads — agents, automation, and high-volume applications. Prepaid balance, pay-as-you-go, no subscriptions or commitments.
- OpenAI-compatible
- Built for high volume
- No lock-in
Eterial Chat
An AI assistant without a monthly subscription. Top up once and chat for months — affordable inference means your money simply lasts longer.
- No subscription
- Top up once
- Lasts for months
Cheap, without sacrificing reliability
Both products run on the same platform. We aggregate compute from multiple capacity sources and route every request to the most cost-efficient one available — with automatic failover to established cloud providers. That's how we keep prices low and keep the service up.
Simple and usage-based
No subscriptions, no lock-in — you only pay for the inference you actually use.
Prepaid balance
You add funds to your account and spend them as you use the service.
Pay only for what you use
Charges are deducted from your prepaid balance based on actual usage. Unused balance stays in your account.
Transparent pricing
Per-model rates will be published on eterial.ai at launch.
Billing in euro
All charges are processed in euro (EUR). No subscriptions, no lock-in.
Built in Europe's most digital-native country
Home of e-Residency, Skype, Wise, and Bolt — and now an AI infrastructure company.
EU jurisdiction
GDPR-grade data protection and European consumer rights by default, not as an afterthought.
Transparent legal environment
Estonian companies operate in one of the world's most transparent business registries — anyone can verify who we are.
Digital-first DNA
A country that runs on digital signatures and online governance is a natural home for an AI infrastructure company.
The principles behind the product
Intelligence should be cheap
The cost of inference is the tax on every AI idea. We exist to lower it.
Pay for what you use
Subscriptions punish light users and cap heavy ones. Prepaid usage-based billing is fair to both.
Reliability is non-negotiable
Cheap compute that's down is worthless. Redundancy and automatic failover are core architecture, not add-ons.
Let's talk
Questions about eterial.ai, the API, or working together? We'd love to hear from you.
savva.riisman@gmail.comTalkingMachines OÜ
Roosikrantsi 8c, Tallinn, 10119, Estonia
Registry code 17503863 · VAT EE102985162