AI Services Schedules

V1.1. Last updated: 3 August 2026

This schedule (“AI Services Schedule”) is incorporated by reference into the Velixo AI Technology Terms and the Velixo Privacy Policy. It identifies the current Model Providers and Velixo AI Gateway services used by Velixo Intelligence, and the applicable data retention periods at each layer. All defined terms have the meanings given in the AI Technology Terms.

Current Model Providers

The following third-party Model Providers are currently engaged by Velixo to process AI Inputs and generate AI Outputs.

Some Model Providers listed below operate as inference gateways that route requests to one or more underlying large language models, rather than operating a single model directly. For these providers, the applicable no-training commitment and data retention terms are those of the specific underlying model handling a given request, as configured by Velixo’s routing settings with that gateway. Velixo restricts such routing to underlying models whose data handling terms are consistent with this Schedule. Where a gateway provider is listed, its entry describes the gateway’s own data handling (e.g., default logging behavior) separately from the terms of the underlying models it may route to.

OpenAI, L.L.C.

• Legal entity: OpenAI, L.L.C., San Francisco, California, United States
• Service: OpenAI API – large language model inference (GPT series)
• No-training commitment: OpenAI does not use API inputs or outputs to train its models
• Abuse-monitoring retention: Up to 30 days – set by OpenAI; cannot specifically be opted out of by Customer or Velixo
• OpenAI Privacy terms: platform.openai.com/docs/models/how-we-use-your-data

Anthropic, PBC

• Legal entity: Anthropic, PBC, San Francisco, California, United States
• Service: Anthropic API – Claude series large language model inference
• No-training commitment: Anthropic does not use API inputs or outputs to train its models
• Abuse-monitoring retention: Up to 30 days – set by Anthropic; cannot specifically be opted out of by Customer or Velixo
• Anthropic Privacy terms: anthropic.com/legal/privacy

Velixo AI Gateway

• Diagnostic log retention: up to 30 days
• Opt-out available: yes, via organization-level administrative controls (see AI Technology AI Terms, section 4)

In addition to the Velixo AI Gateway layer operated by Velixo and the Model Providers listed above, Velixo engages the following subprocessors to support the operation of Velixo Intelligence:

OpenRouter, Inc.

• Legal entity: OpenRouter, Inc., New York, New York, United States
• Service: OpenRouter API – inference gateway/router providing access to multiple third-party large language models
• Scope: OpenRouter does not itself train or operate the underlying models it routes requests to. Data handling, retention, and training commitments are governed by the policies of whichever downstream Model Provider handles a given request, as selected by Velixo
• OpenRouter-level logging: Prompt and completion content is not logged by OpenRouter
• No-training commitment: Depends on the downstream Model Provider selected for each request. Velixo restricts routing to providers with no-training and abuse-monitoring terms consistent with this Schedule
• Abuse-monitoring retention: Varies by downstream Model Provider; see that provider’s entry in this Schedule where applicable
• OpenRouter Privacy terms: openrouter.ai/privacy and  openrouter.ai/terms

Langfuse (operated by ClickHouse, Inc.)

• Legal entity: ClickHouse, Inc., Redwood City, California, United States
• Service: Langfuse Cloud (US region – Oregon, AWS us-west-2) – LLM observability and prompt management platform
• Purpose: (1) Quality, cost, and performance analysis of AI Inputs and AI Outputs; (2) Prompt management
• No-training commitment: Langfuse does not train or fine-tune ML/LLM models on Client Data
• Retention: AI Inputs and AI Outputs sent to Langfuse are retained for up to 30 days, as configured by Velixo, after which they are purged
• Langfuse/ClickHouse Privacy terms: langfuse.com/privacy/privacy (subprocessor list:
  langfuse.com/security/subprocessors)

Updates to This Schedule

Velixo will update this Schedule when Model Providers are added, replaced, or removed, or when retention periods change. Customers will be notified by email or in-product notification. Continued use of Velixo Intelligence after an update takes effect constitutes acceptance of the revised Schedule.