Draft pending legal review
This document is a draft and has not yet been approved by counsel. It is provided for transparency and may change before it becomes binding.
Terms of Service
Placeholder legal copy — replace with counsel-reviewed text before production.
# Terms of Service
**Last updated:** 2026-07-03
**Status:** Draft pending legal review.
These terms govern your use of the Graphene platform: the consumer portal, APIs, and related services (the "Service"), operated by Tokenomik Australia Pty Ltd ("Graphene"). By creating an account or using the API you agree to these terms. Enterprise customers may have a separate agreement; where it conflicts with these terms, the agreement wins.
## 1. The Service
Graphene provides an orchestration layer that routes AI inference requests to third-party model providers and GPU capacity through a single OpenAI-compatible API, with metering, billing, budgets and routing policy. We do not create the models. Model outputs are generated by the model you select; outputs may be inaccurate and you are responsible for evaluating them before relying on them.
## 2. Accounts and workspaces
You are responsible for your account, your workspace members, and all activity under your API keys, including activity by automated systems and agents you deploy. Keep keys secret; rotate them if compromised.
## 3. Fees and billing
Usage is billed per token at the rates shown in the dashboard at the time of the request. Published rates reflect a discount to each model provider's list price and may change as list prices change. Budgets and alerts are controls, not guarantees; you remain responsible for charges incurred before a budget takes effect. Taxes are added where applicable. Payments are processed by Stripe.
## 4. Acceptable use
You must not use the Service to violate law, to infringe others' rights, to attempt to extract or attack the models or infrastructure, to develop weapons, to generate CSAM or content that exploits minors, or to resell raw access in breach of upstream model providers' terms. Your use of specific models is also subject to the applicable model provider's usage policies, which we pass through. We may suspend keys or accounts for breach, abuse signals, or non-payment.
## 5. Your content
You retain your rights in prompts and outputs to the maximum extent the upstream model provider's terms allow. You grant us the limited rights needed to operate the Service: to transmit, cache, meter and route your content per your configuration. We do not use your content to train models.
## 6. Routing, residency and service levels
Routing decisions are made by the platform per your configuration and recorded in an auditable ledger. Residency and sovereign-routing commitments, uptime commitments, and support terms apply only where stated in an enterprise agreement. The self-serve Service is provided as-is and as-available.
## 7. Automated and agent access
The Service may be accessed by automated systems using API keys you control. Where we offer machine-readable capability manifests, benchmark endpoints or protocol-based settlement (such as x402) in preview, those surfaces are experimental, may change without notice, and are excluded from any service commitments until covered by an agreement.
## 8. Liability
To the maximum extent permitted by law (including the Australian Consumer Law, whose non-excludable guarantees are not affected), our aggregate liability is limited to the amounts you paid us in the 3 months before the event, and neither party is liable for indirect or consequential loss.
## 9. Termination
You may stop using the Service at any time; export tooling and OpenAI compatibility mean migration is a base-URL change. We may terminate for breach with notice where practicable. Sections that by nature survive (fees owed, liability, content rights) survive termination.
## 10. Governing law
These terms are governed by the laws of Victoria, Australia, and the parties submit to the jurisdiction of its courts.
## 11. Contact
Questions about these terms: use the Contact page or your account representative.