Model Providers

What this page is

LemonSugar Ai is a routing and chat product. We do not train our own foundation models. When you send a prompt, we select a model and forward the request to one of the third-party providers below — either directly to the provider's API, or through OpenRouter. Each provider's own terms, usage policy, and privacy policy apply to the content of your request and the model's output.

The full list of vendors that process personal data on our behalf (database, payments, email, etc.) lives on the Sub-processors page. This page is narrower: it covers only the model providers that receive prompt and output content.

Providers

Provider terms flow down to you

By using LemonSugar Ai you agree to comply with each model provider's terms and acceptable-use policy for any request routed to that provider. We may suspend or terminate your access to a model — or, in serious cases, to the Service — if your use would breach an upstream provider's policy or expose us to liability under our agreements with them.

We do not control which provider receives a given request: the router picks the most cost-effective model for the task. If you need a request to stay on a specific provider, pin the model in Settings → Models, or bring your own key (BYOK) so the request goes directly to that provider.

Prohibited across all providers

The specifics vary, but every model provider we route to forbids at least the following categories. These are prohibited on LemonSugar Ai regardless of which model you choose:

• Child sexual abuse material (CSAM) or any sexual content involving minors.
• Content that incites violence, terrorism, or mass harm, or that promotes self-harm or suicide.
• Generation of weapons of mass destruction, including biological, chemical, radiological, nuclear, or cyber-weapon capabilities.
• Non-consensual intimate imagery, sexual deepfakes, or any sexual content depicting a real identifiable person without consent.
• Targeted harassment, hate speech, or content that dehumanizes a protected group.
• Fraud, scams, malware, credential theft, or instructions to circumvent security controls.
• Unauthorized practice of regulated professions (medical, legal, financial advice presented as licensed) and election or political-process manipulation.
• Mass surveillance, biometric identification of private individuals, and social-scoring systems prohibited under EU AI Act Article 5.
• Attempts to extract model weights, reverse-engineer system prompts, or red-team production endpoints without written approval.

Our full house rules — including how we respond to reports — are in the Responsible Ai policy. The contractual basis is in the Terms.

Data handling

Prompt and output content is sent to whichever provider serves the request. We have configured each provider account to opt out of training on customer data where the provider offers that setting (OpenAI Business, Anthropic Commercial, Google Vertex). OpenRouter and individual open-source hosts have their own retention windows — see their privacy policies linked above.

Data residency, retention, and deletion commitments for enterprise customers are in the Data Processing Addendum.

Changes

We add, remove, and re-rank model providers as the market changes. Material additions or replacements are announced with at least 30 days' notice to enterprise / DPA customers. To subscribe, email hello@lemonsugar.ai with the subject “Subscribe to model-provider updates”.