OpenAI · Model provider
Can we use OpenAI models and still control keys, cost and what is sent?
Yes. You add your own OpenAI API key to the Katonic AI Gateway, and agents call OpenAI models through the gateway. The models run in OpenAI's cloud, so prompts and responses go to OpenAI. The gateway holds the key, decides which teams may call which model, applies rate limits and budgets, checks each turn against your guardrail profile and records usage and cost.
Where the model runs.
In the provider's cloud.
How it is connected
- 01Your OpenAI API key.
- 02The address is filled in for you.
In OpenAI's cloud.
What leaves your environment
The prompt and the response.The prompt and the response go to OpenAI. Katonic controls which teams and models may do that, checks the turn against your guardrail profile and records it. In an install flagged as air-gapped, hosted providers start disabled.
Keys
Your key is kept encrypted and decrypted only for the call. It is not stored or cached in plain text.
Model access
A policy for each tenant and an entitlement for each team decide which models may be called.
Rate limits
Requests and tokens per minute, for each user, tenant and model.
Budgets
Checked before the call, at three levels: warn, soft and hard. A hard budget blocks the call.
Guardrails
The input and output of a chat turn are checked against your guardrail profile.
Failover
Health is tracked for each provider and model, with a circuit breaker. A call can move to the next member of its tier.
Cost and usage
Each call is priced when it is recorded, and usage and cost can be queried.
Who paid
Each usage record says whose credential paid for the call.
Air gap
In an install flagged as air-gapped, hosted providers start disabled.
A model you serve on your own GPUs keeps the prompt and the response on your infrastructure. The gateway does the same work in front of it.
Does our data go to OpenAI?
Yes. The prompt and the response of a call to an OpenAI model go to OpenAI. To keep them inside, call a model you serve yourself.
Can we allow OpenAI for some teams and not others?
Yes. A policy for each tenant and an entitlement for each team decide which models may be called.
What happens in an air-gapped install?
Hosted providers start disabled.