KEYVERA provides a unified interface to supported third-party AI models. This page identifies important provider-policy sources and explains how provider-specific requirements interact with the KEYVERA Terms of Service.
KEYVERA may obtain model access through upstream platforms, aggregators, infrastructure providers, or other authorised channels rather than through a direct contract with the model developer. A provider name or trademark does not imply partnership, sponsorship, endorsement, or direct contractual privity.
What customers should know
- The KEYVERA Terms govern your relationship with KEYVERA.
- Additional provider rules may apply to a selected model or capability.
- Data handling, retention, location, and safety controls can differ between models and routes.
- Provider availability and model versions can change without a guaranteed migration period.
- Review the current Account or model catalogue before regulated or sensitive use.
- Provider names and marks belong to their respective owners.
1. Relationship and contract hierarchy
Your contract for the KEYVERA Services is with KEYVERA B.V. under the Terms of Service and applicable Order. Provider policies identified for a model form part of the usage restrictions only to the extent disclosed by KEYVERA and applicable to the relevant Service.
Provider pages linked below are maintained by independent third parties and may change without notice. If this page conflicts with a restriction clearly displayed in the Account, Documentation, or purchase flow for a particular model, the more specific disclosed restriction controls for that model, subject to mandatory law.
2. Provider policy references
These links are provided for transparency and convenient reference. The exact model catalogue and applicable route can change, and not every document linked below necessarily forms part of every KEYVERA Order.
OpenAI
GPT and related models, where offered. Review current business, service, and usage requirements before deployment.
Anthropic
Claude models, where offered. Commercial and usage requirements may differ from consumer-facing Claude terms.
Google Gemini
Gemini models, where offered. Google AI Studio, Gemini API, Vertex AI, and upstream routes may have different terms.
DeepSeek
DeepSeek models, where offered. Open-platform terms contain requirements for downstream applications and end users.
xAI
Grok models, where offered. Developer and business API use is governed by enterprise-facing requirements.
Z.AI / Zhipu AI
GLM and related models, where offered. Regional platforms and specific plans may impose different restrictions.
External links may be updated or redirected by the provider. KEYVERA does not control third-party pages. Customers should retain the policy version relevant to a regulated deployment or material production decision.
3. Data handling and model selection
Provider-specific handling may differ with respect to processing location, retention, human review, abuse monitoring, model improvement, zero-retention eligibility, and support for regulated data. A brand name alone is not enough to determine those characteristics because the route, account type, configuration, and upstream channel may matter.
Do not submit regulated, highly sensitive, or confidential information until you have verified the current model information, the KEYVERA Privacy Policy, and any required DPA or written assurance.
4. Safety filters and blocked requests
Providers or upstream channels may apply safety classifiers, content filters, geographic restrictions, rate limits, or abuse controls. As a result, similar prompts may produce different results or be blocked across models. KEYVERA does not guarantee that a particular request will be accepted or that provider controls can be disabled.
You must also comply with the KEYVERA Acceptable Use Policy. The absence of a provider-side block does not make an otherwise prohibited use permissible.
5. Model and provider changes
Providers may rename, update, deprecate, suspend, or withdraw models. KEYVERA may change routing, supported versions, or availability to maintain security, compliance, quality, or continuity. Where reasonably practicable, KEYVERA will provide notice of material changes affecting paid Services, but emergency or upstream changes may take effect immediately.
Customers should use model identifiers, fallbacks, testing, and monitoring appropriate to the importance of their application.
6. Questions and current applicability
For a production or regulated use case, ask KEYVERA which provider, route, policies, and data-handling terms apply before submitting data. Contact legal@keyvera.cloud and identify the model and intended use.
Last verified: . Provider pages can change after this date.