Power AI employees / AAES Infer
Qualification requiredAAES Infer brings model access and spending under company control. Run Infer in your own environment or let AAES host it. Use your own provider accounts or buy model access through AAES, with a qualified processing jurisdiction and explicit data-use terms.
Choose your model access / AAES Infer
Model access gets harder to manage when each application has its own provider account, billing rules and data settings. Infer gives your team a defined offer to approve before sending customer data.
Select an available model and country combination. The routing engine refuses requests outside the configured jurisdiction. It does not automatically fall back to another country or model.
An EU boundary may include several EU countries. A requirement for one EU member state needs a separate offer for that country. The UK and Switzerland are separate markets.
Each offer must prohibit using prompts and outputs for training, fine-tuning or model improvement, including reuse across customers. Retention, safety processing and any permitted access by subprocessors must be disclosed before you buy.
Availability depends on the provider accepting those terms and the required account settings being verified.
Your company manages payment terms and project spending limits centrally. Employees and agents use authorized project access without making individual purchases. A prepaid company balance is shared across configured projects and qualified models within each jurisdiction. Administrators use their existing AAES login.
For prepaid AAES access, a maximum request charge is reserved against paid funds before a provider is contacted. With your own provider account, the reservation uses your company usage budget. Project limits apply in both cases, and usage settles from the provider’s reported token counts.
Usage billing depends on the qualified provider and offer. Dedicated open-model capacity needs a separate quote for the model, location, reserved capacity, minimum term, setup and operating costs. Capacity is ordered only after agreement and advance funding.
Run the gateway and its controls on infrastructure you manage. Connect your own provider accounts, or use project-scoped AAES access. When you buy through AAES, calls also pass through an AAES gateway qualified for the same jurisdiction. AAES keeps payment and credit authority on that gateway.
Your infrastructure, provider route, logs, storage and support arrangements must meet the offer’s residency and privacy requirements.
AAES operates the gateway for your company. You can connect company-owned provider accounts or buy access through AAES. The hosting location and provider processing locations must both meet your selected jurisdiction.
The AAES-hosted deployment is prepared for AWS. Customer content connects directly to a regional AWS gateway, and every model destination must qualify for the selected jurisdiction. Live availability depends on account, infrastructure and provider acceptance.
The provider bills your company directly for model usage. Administrators set a company usage budget and project limits in Infer. These limits do not create an AAES cash balance or credit account. Your provider account still needs verified privacy settings and a qualified local route.
Project credentials grant access within your company’s spending limits. A customer-hosted installation never receives AAES’s reusable provider credentials.
Use your own model-provider accounts, with usage billed directly by the provider, or buy access through AAES. AAES company billing includes prepaid funding and approved invoice accounts. Invoice-funded inference is not available in the current release. All live inference requires separate activation.
Use text chat, native Messages or native Responses interfaces, with streaming and client tools where the signed offer supports them. Select a named model version and jurisdiction. Only a qualified local offer can serve requests; each offer lists the API features it supports. Read the Infer API scope, required headers and request limits.
AAES Operate runs agents. AAES Govern checks supported actions routed through it. Infer is a separate model-access service. Infer includes streaming Messages and Responses interfaces for native Claude and Codex clients, including supported client tool calls and results. Direct and governed connections require qualification of the exact provider route and acceptance of the deployment.
Govern’s existing model-source integrations describe gateway connectivity. They do not establish availability through Infer.
Check availability / AAES Infer
Choose the processing jurisdiction and model version your company requires. An offer identifies the provider, approved destinations, data terms and price. A catalog entry alone does not establish availability.
Access opens only after provider terms, the exact deployment, billing arrangements and production acceptance have been confirmed for your offer. Contact us to assess your requirements. Supplier partnerships and service-level commitments must be stated in the offer.
These are target markets, not live locations. Model availability differs by country. A provider must establish model access, processing location and the required data terms for each offer. We cannot promise every model in every market.
An invoice account requires company approval, an agreed credit limit and payment terms. Invoice-funded inference is not available in this release. See billing and activation status.
Pricing is not published yet. Usage billing will depend on upstream rates and AAES service costs. Dedicated capacity may lower unit costs at sustained usage, but it carries a minimum commitment. We will quote the model, location and usage assumptions before you pay.
An offer must state where inference, content storage, logs and backups are processed. Support access, account metadata and legally required disclosures need their own terms. A country-specific API address alone does not establish these guarantees.
Infer offers require a prohibition on training and secondary reuse of your prompts and outputs. An offer can only be activated after its contractual terms and technical controls have been verified. The selected provider must process your request to return an answer. Its authorized subprocessors and necessary service processing must be disclosed. The strict profile requires zero retained prompt/output content and prohibits training, evaluation and product-improvement reuse; a route that cannot meet it stays unavailable.
Public access is not available. Contact us with the model, processing country and expected usage you need so we can assess whether a qualifying offer is possible.
Scope your offer / AAES Infer
Include your required country, expected usage and data handling requirements.