Last updated: 3 August 2026
AI Service Disclosure
Model Gate provides technical access to third-party artificial-intelligence models. Model Gate is not the developer or owner of most models offered through the Service.
Upstream Providers
Depending on model selection and availability, processing may involve OpenAI, Anthropic, Google, xAI, Mistral AI, DeepSeek, Cohere, Together AI, Fireworks AI, Alibaba Cloud/Qwen or another disclosed provider.
Routing
A request may be routed based on the selected model, provider availability, reliability, latency, geographic constraints, technical compatibility and commercial configuration.
Data Handling
- Synchronous prompt and response content is not intentionally retained after transient processing.
- Asynchronous prompt and response content is retained for approximately one hour.
- Model Gate does not use prompts or responses to train Model Gate models.
- Model Gate does not intentionally send the customer email, Model Gate user ID or originating IP address as part of the upstream model request.
- No separate “zero retention” endpoint is currently offered.
Output Limitations
AI output is probabilistic and may contain factual errors, fabricated references, bias, unsafe content or material resembling third-party content. Output must be reviewed before use.
Professional and High-Risk Use
AI output is not a substitute for qualified professional judgement. Customers are responsible for appropriate human review and compliance when using AI in regulated or high-impact contexts.
Provider Changes
Models, providers, versions, context limits, output limits, safeguards and prices may change. Model Gate may replace or discontinue a route where required for security, legality, reliability or commercial availability.