Four ways to run AI on your data.

Exalt Reserve

Three other ways to use AI with company data: a hosted AI API, a build of your own, or another Exalt deployment. Each fits different work.

Reserve or a hosted AI API

Where the models run
A hosted AI API runs the provider's models on the provider's infrastructure. Reserve runs open models in your own cloud account.
Which models
Hosted APIs offer the provider's own models, including frontier models. Reserve self-hosts Llama, Mistral, Qwen, Gemma, and DeepSeek. Which models run is decided with you for each use case.
How your software calls it
Both are called from your software through an API. Reserve's has no public endpoint; your applications reach it from your private network.
Whose terms apply
What you send to a hosted API is handled under the provider's terms. With Reserve, questions go by default to models in your own account.
Who you pay
With a hosted API, you pay the provider, usually by usage. With Reserve, Exalt bills for its software, with no per-prompt or token usage fees, and you pay your cloud provider directly for the infrastructure.
When the API fits better
When the work needs a frontier model. Whether a Reserve use case also uses an outside AI service is decided with you, and Exalt Hybrid is built to route approved tasks to an approved frontier API.

Build: Your team picks the open models, deploys them, and builds retrieval, agents, and monitoring around them.
Reserve: Exalt deploys the platform, with models, retrieval, agents, and monitoring, into your account.
Both: The account, the bill, and the data stay yours.