Pricing Use cases Data Sign in Get started
Legal

AI Governance Policy

How we run, source and oversee the AI models behind the Tuas AI Model API, and the commitments we make about your data.

Effective 4 November 2026

1. Our commitments

  • Our own hardware, in Singapore. We use only certified, physical NVIDIA GPU accelerators, on premises, owned and operated by Tuas AI. Inference does not run on shared public-cloud GPU capacity.
  • No retention. Prompts, context and outputs are processed for the duration of the session only and are not kept afterwards.
  • No training. We never train or fine-tune any model on session data, prompts, context or outputs.
  • Content stays with us. Customer content does not leave our infrastructure and is never sent to the model vendor.

2. Model sourcing

Models are delivered and distributed in flight, with Hugging Face as our model weight provider. We deploy the weights onto our own infrastructure and serve them from there. Only model weights flow from Hugging Face to us; no customer data flows back.

Before we offer a model, we check its source, licence and published documentation, and test it for quality and stability. We follow each model’s licence and acceptable-use terms and identify the model and its developer on our website.

3. Security and access

Access to production systems is limited to authorised staff, on a need-to-know basis. Traffic to the API is encrypted in transit. We monitor service metadata (not content) to keep the Service secure and to detect abuse.

4. Responsible use

Customers decide how they use model outputs and remain responsible for their own use cases, including human review of decisions that significantly affect people. Our Terms of Use set out prohibited uses. Report misuse to [email protected]; we investigate reports and may suspend access.

5. Transparency

We publish which models we serve, who developed them and what they cost. AI outputs can be wrong; we say so plainly and encourage customers to tell their own users when they are interacting with AI.

6. Frameworks

This policy is informed by Singapore’s Model AI Governance Framework and the Personal Data Protection Act 2012. Our handling of personal data is described in our Privacy Policy.

7. Incidents and review

We investigate security and model incidents promptly and notify affected customers and authorities where required. We review this policy at least once a year and when we materially change our models or infrastructure.

Who we are

Tuas AI (UEN 202635139W) is a company registered in Singapore. Registered address: 1 Kaki Bukit View, #02-10, Techview, Singapore 415941. Email [email protected] for privacy and data protection matters, or [email protected] to report misuse. We do not offer support by phone.

Model API

PricingUse casesData and privacy

Legal

Privacy PolicyTerms of UseAI Governance Policy

Contact

[email protected][email protected]

Tuas AI Model API, hosted in Singapore. Live now. Prices in USD per 1M tokens, as at September 2026.
© 2026 Tuas AI, a Singapore company (UEN: 202635139W). 1 Kaki Bukit View, #02-10, Techview, Singapore 415941.