AI Services · 04
Inference, measured by the token.
Run leading open-weight and domain models through an API that works with the client libraries your team already uses — and pay for the tokens you use.
Overview
AI Token & Inference API
Most teams want the benefits of large language models without operating GPU fleets themselves. Our inference service gives you reliable model access, metered by the token, from infrastructure we run.
For sensitive workloads, the same models can be deployed privately, in an environment dedicated to you.
What we offer
Token-metered API
Simple, usage-based access to text and multimodal models.
Compatible interface
Works with widely used client libraries and tooling.
Open-weight and domain models
A curated catalogue, including industry-specific models.
Private deployment
Dedicated endpoints or deployment into your own environment.
Your data stays yours
Prompts and outputs are not used to train models without your consent.
Usage visibility
Clear reporting on requests, tokens and spend.
How it works
- 01
Choose models
We help you pick models for your use case and budget.
- 02
Connect
Integrate through the API with your existing code.
- 03
Evaluate
Test quality, latency and cost on your own data.
- 04
Scale
Move to production with capacity planned ahead.
Compliance & safeguards
Built in, not bolted on.
Customer prompts and outputs are not used for training without consent.
Responsible AI →Requests are encrypted in transit and access is logged.
Data Security →Use of the API is governed by our Acceptable Use Policy.
Acceptable Use Policy →