Terms of Service

By using our Services, you acknowledge and agree that:

Marketing Website (aion.xyz):

• CookieYes for GDPR-compliant consent management
• Google Tag Manager and Analytics 4 (with your explicit consent)
• Meta Pixel, LinkedIn Insight, and Twitter/X tracking (with your explicit consent)
• You may browse without accepting marketing cookies

Platform, Inference, and Agent Services (console.aion.xyz, api.aion.xyz):

• PostHog for product analytics (REQUIRED – cannot opt out while using the platform)
• Stripe for payment processing (REQUIRED for billing operations)
• Intercom for customer support (REQUIRED for assistance)
• These are essential services for platform operation
• By using the platform, inference, or agent Services, you consent to these essential data processors

We reserve the right to modify these Terms, our Privacy Policy, Cookie Policy, Acceptable Use Policy, and any associated policies. Modifications may be made for reasons including but not limited to:

• Evolving business requirements or strategy
• Regulatory compliance updates (including evolving AI regulation)
• Technology, model, or infrastructure changes
• Service enhancements or restrictions
• Integration of new features, models, or third-party services

Notice will be provided at least 30 days in advance for material changes affecting your rights or obligations, via email to your registered address or platform notification. Continued use after notice constitutes acceptance of the modifications. If you disagree with any changes, you must discontinue use before they take effect.

These Terms constitute a binding legal agreement. You must accept them in their entirety. By accessing or using any part of our Services, you:

• Acknowledge you have read and understood these Terms completely
• Agree to be bound by all provisions
• Consent to our data collection and processing practices as described in our Privacy Policy
• Accept that non-agreement requires cessation of all use

Where you and AION have signed a separate written agreement (for example, an enterprise master services agreement, order form, or data processing agreement), that agreement governs and controls over any conflicting provision in these Terms.

• GPU Instance Provisioning: bare-metal and managed GPU instances
• Managed Infrastructure: pre-configured environments for AI/ML workloads
• Resource Management: project-based allocation and quota management
• API Access: programmatic control of infrastructure
• Monitoring and Metrics: real-time resource utilization tracking

• Access via API to models hosted and served on AION-operated infrastructure for inference (“Inference Services”)
• Supported modalities are text, vision (including vision-language models), and embeddings. Additional modalities such as voice may be offered in the future.
• Inference is operated on a zero-retention basis: your Inputs and Outputs are processed only to serve your request and are not stored or logged by AION (see Section 5)
• Usage-based access subject to rate limits, quotas, and context/throughput constraints
• Availability of specific models is subject to change; models may be added, deprecated, or removed with reasonable notice where practicable

AI agents and agentic workflows ("Agent Services") are not offered on a self-serve basis through the platform. They are delivered to enterprise customers under a separate written agreement. Where an agent solution is deployed within your own cloud environment (for example, your VPC), your data remains in that environment and does not leave your cloud; whether it is logged or used to train is determined by your own configuration and the separate agreement, not by these Terms. The provisions of Sections 4 and 6 concerning acceptable use and human oversight apply to any agent solution you operate.

Fine-tuning and custom model hosting ("Custom Models") are not currently offered on a self-serve basis through the platform. Where AION provides fine-tuning or custom model services under a separate written agreement, then, as between you and AION and subject to any applicable third-party base-model license, you retain rights in Custom Models trained on your data. Some third-party or open-weight base-model licenses may restrict rights in derivative models; those license terms will govern in that case.

• Provisioning Time: target of <20 minutes for GPU instances
• Availability: best-effort availability with no guaranteed uptime except where a separate SLA has been agreed in writing
• Maintenance: regular maintenance windows with advance notification where practicable
• Model and Regional Availability: subject to provider capacity, model availability, and regional restrictions

To use the Services, you must:

• Provide accurate and complete registration information
• Be at least 18 years old (or 16 with parental consent in the EU) and have legal capacity to enter contracts
• Maintain the security of your account credentials
• Comply with all applicable laws and regulations
• Pass our identity verification requirements

For organization accounts:

• The registering party must have authority to bind the organization
• You are responsible for all users under your organization
• Project-level and role-based access controls must be properly managed
• Billing responsibility lies with the organization owner

You are responsible for:

• Maintaining strong, unique passwords and enabling multi-factor authentication
• Protecting API keys, access tokens, and any credentials you provide to agents or integrations
• Immediately reporting any unauthorized access
• All activities conducted under your account, keys, and agents

You may not use the Services to:

• Conduct illegal activities or violate any applicable law
• Mine cryptocurrency without explicit written permission
• Distribute malware, viruses, exploits, or harmful code
• Perform unauthorized penetration testing, security scanning, or attacks
• Infringe intellectual property or privacy rights
• Harm our infrastructure, models, or other users
• Resell or sublicense resources or model access without authorization
• Circumvent usage limits, quotas, rate limits, or safety systems
• Operate open proxies, anonymous VPN services, or denial-of-service tooling

You may not use the Inference Services, Agent Services, Custom Models, or any Outputs to create, facilitate, or distribute:

• Child sexual abuse material (CSAM) or any content that sexualizes minors
• Non-consensual intimate imagery, or sexualized or defamatory synthetic media (“deepfakes”) of real individuals
• Content that incites, facilitates, or provides instructions for violence, terrorism, self-harm, or the development of weapons, including chemical, biological, radiological, nuclear, or high-yield explosive weapons
• Harassment, bullying, hateful content, or unlawful discrimination
• Fraud, scams, phishing, deceptive impersonation, synthetic identities, or malicious social engineering
• Political disinformation, election manipulation, or coordinated inauthentic behavior
• Synthetic media of real people or AI-generated content in contexts requiring disclosure, without providing that disclosure and obtaining any required consent
• Unauthorized biometric identification, mass surveillance, social scoring, or emotion inference in prohibited contexts

You may not use the Services to make fully automated decisions that produce legal or similarly significant effects on individuals — including in employment, credit, housing, insurance, education, healthcare, essential services, or legal matters — without meaningful human oversight and any disclosures, assessments, or safeguards required by applicable law. You are the deployer of any such system and are responsible for its lawful, safe, and non-discriminatory operation.

Where the EU AI Act or comparable AI legislation applies to your use, you are responsible for determining the risk classification of your use case and meeting the corresponding obligations (including prohibited-practice restrictions, transparency for AI interactions and synthetic content, and any high-risk obligations). The Services are general-purpose tools; you are responsible for how you deploy them.

You may not use the Services or any Outputs to train, fine-tune, or improve a competing foundation model, nor attempt to extract, reverse-engineer, or reconstruct model weights, training data, or system prompts, except to the extent this restriction is prohibited by applicable law or expressly permitted by a base model's license.

"Inputs" means the prompts, files, data, credentials, and other content you or your agents submit to the Services. "Outputs" means the content the Services generate in response to Inputs. "Customer Content" means Inputs, Outputs, and other data you process through the Services.

As between you and AION, you own your Inputs. To the extent permitted by applicable law, AION assigns to you all of its right, title, and interest in Outputs generated for you. You are responsible for your Customer Content and for ensuring you have the rights and any consents necessary to submit it and to use the Outputs.

You grant AION a limited, non-exclusive license to host, process, transmit, and display Customer Content solely as necessary to provide, secure, and maintain the Services for you, and to comply with law. We do not claim ownership of your Customer Content.

We do not use your Inputs, Outputs, or other Customer Content to train, fine-tune, or improve any models — whether our own or those of third parties. The Inference Services operate on a zero-retention basis: your Inputs and Outputs are processed only to serve your request and are not stored or logged by AION, including for abuse detection, safety filtering, or debugging. We may generate and use aggregated, de-identified operational metadata (such as latency, error rates, and volume) that does not include Customer Content to secure, monitor, and improve platform performance. Where you engage AION for fine-tuning or a custom deployment under a separate written agreement, that agreement governs any use of your data for training, which occurs only at your direction.

Given the nature of machine learning, Outputs may not be unique, and the Services may generate the same or similar Outputs for you and for other customers. AION makes no representation that Outputs are original or that your use of Outputs will not infringe the rights of third parties. You are responsible for reviewing Outputs before relying on or distributing them.

The Services may provide access to models released under open-weight, community, or research licenses. These models are hosted and served on AION-operated infrastructure; your Inputs and Outputs are not sent to, or shared with, any third-party model provider. Your use of such models and their Outputs is nonetheless subject to the applicable model's license and acceptable-use terms, and you are responsible for compliance with them. We will make the applicable terms available or identify the relevant model on request.

Outputs are generated by probabilistic AI systems and may be inaccurate, incomplete, outdated, or offensive, and may not reflect current facts. The Services may "hallucinate" — produce plausible but incorrect content. Outputs do not constitute professional advice (including legal, medical, financial, or safety advice) and must not be relied upon as such.

You are responsible for evaluating Outputs for accuracy and suitability for your use case, and for implementing appropriate human review before relying on Outputs in any consequential context.

Where you deploy Agent Services, agents may take autonomous actions — including calling tools and reading from or writing to systems you authorize. You are responsible for the scope, permissions, guardrails, and monitoring you configure, and for all actions an agent takes within the access you grant, including actions in connected third-party systems. AION is not responsible for actions taken by an agent within the scope you authorize, or for the consequences of insufficient oversight, permissions, or guardrails on your part.

• Payment Processor: all payments are processed through Stripe
• By providing payment information, you authorize us to charge your payment method
• Payment data is handled directly by Stripe and not stored on our servers

• Compute: GPU instances are billed per hour or fraction thereof
• Inference: billed on a per-token basis, as described in your plan or order
• Agents and custom deployments: charges are set out in the applicable separate written agreement
• Usage Tracking: real-time usage tracking via the platform dashboard
• Price Changes: we may modify pricing with 30 days’ notice
• Taxes: you are responsible for all applicable taxes

For current billing models, payment terms, and pricing structures, contact support via Intercom, email [email protected], or review docs.aion.xyz/billing.

• Disputes must be reported within 30 days of the charge
• We will investigate and respond within 10 business days
• Refunds are at our sole discretion

Your use of the Services is governed by our Privacy Policy and Cookie Policy. By using the Services, you consent to the data practices described there.

We use PostHog for product analytics on console.aion.xyz, which may include session recording, behavior tracking, feature usage, and performance monitoring. This tracking is essential for platform operation and cannot be disabled while using the platform. It does not extend to the contents of your Inputs and Outputs processed through the Inference or Agent Services, which are handled as described in Section 5 and the Privacy Policy.

For enterprise customers requiring a Data Processing Agreement, standard DPAs incorporating EU Standard Contractual Clauses are available on request and cover our sub-processors. Contact [email protected].

All Services components remain our exclusive property, including software, code, model architectures and weights we own, user interfaces, documentation, trademarks, and APIs. Nothing in these Terms transfers ownership of the Services or underlying models to you.

We grant you a limited, non-exclusive, non-transferable license to access and use the Services for their intended purposes and in accordance with our documentation.

Any feedback you provide may be used by us without restriction or compensation and does not create confidentiality obligations.

THE SERVICES, INCLUDING ALL MODELS, INFERENCE, AGENTS, AND OUTPUTS, ARE PROVIDED "AS IS" AND "AS AVAILABLE" WITHOUT WARRANTIES OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, NON-INFRINGEMENT, ACCURACY, AVAILABILITY OR UPTIME, AND ERROR-FREE OR UNINTERRUPTED OPERATION.

TO THE MAXIMUM EXTENT PERMITTED BY LAW:

• We are not liable for indirect, incidental, special, or consequential damages
• Our total aggregate liability is limited to the fees you paid in the 12 months preceding the claim, except where this limitation is prohibited by law (including for data protection violations under GDPR)
• We are not responsible for data loss or corruption, for the accuracy or consequences of Outputs, or for actions taken by agents within the scope you authorize
• We are not liable for third-party actions, models, or content

We may modify or discontinue features, change specifications, deprecate or remove models, and implement new restrictions or quotas, with reasonable notice where practicable.

You may terminate your account through the dashboard or by contacting support. Termination does not relieve payment obligations.

We may terminate or suspend your account for Terms violations, non-payment, suspicious or fraudulent activity, misuse of the Inference or Agent Services, or as otherwise permitted with notice.

Upon termination, access ceases, running instances and agents are terminated, Customer Content may be deleted after the recovery period described in the Privacy Policy, outstanding charges remain due, and licenses granted terminate.

These Terms are governed by the laws of Delaware, United States, without regard to conflict-of-law principles. For EU residents, mandatory consumer protection laws of your country of residence apply and cannot be overridden by this choice of law.

Any disputes shall be resolved through binding arbitration administered by JAMS under its Streamlined Arbitration Rules, seated in Delaware (unless the parties agree otherwise), on an individual basis only (no class actions), with limited discovery and expedited resolution.

Small claims court actions, injunctive relief for IP violations, and claims related to unauthorized access are not subject to arbitration.

You represent that you are not located in an embargoed country, are not on any prohibited-party list, will comply with all export control laws, and will not use the Services for prohibited end uses.

The Services are subject to U.S. export controls, which may include controls applicable to advanced computing, AI models, and model weights. You may not export or re-export the Services, any model, or model weights without appropriate authorization, and you are responsible for compliance with export controls applicable to your use, including for dual-use and controlled end uses.