AI Gateway: give each team its own models, budgets and guardrails in Mission Control

AI Gateway lets you give every team in your company inference access without sharing a master API key. In Mission Control, create a gateway per team, choose its models, set its budget and guardrails, and issue each person a scoped inference key. AI Gateway is free. You pay for the inference that runs through it, and Telnyx-hosted models run with zero data retention.

What's new

• Gateways per team: Set the models a team can use, across Telnyx-hosted models and your own Anthropic and OpenAI keys.
• Rules: Cap spend per gateway, person, key or end user over 1, 7 or 30 days, with requests and tokens per minute limits.
• Guardrails: Flag or block secrets, card numbers, IBANs, SSNs, emails and phone numbers, separately on prompt and response.
• Inference keys per person: Issue a key that narrows the gateway's models to a subset, with expiry and revocation.
• Usage reporting: See spend by gateway, person, key or end user, plus every guardrail flag and block.
• Full API: Every portal action is available through the management API.

Why it matters

A single shared API key hands every model and every budget to anyone who holds it. AI Gateway reserves cost before each request runs, so a burst of parallel calls cannot overshoot a cap, and guardrails flag or block secrets and sensitive data before a prompt ever reaches a model. Telnyx-hosted models add zero data retention, and moving a team to open-source models on GPUs Telnyx owns, including NVIDIA B300s, is a change to one list.

Example use cases

A company might give Engineering access to every Telnyx-hosted and BYOK model with a $1,000 budget every 7 days, and block secrets in prompts so nobody pastes a credential into a model. Its Support team could use Telnyx-hosted open-source models only, with any response that contains an IBAN or a card number blocked before it comes back. Teams building AI into their own product can also cap each of their customers. Pass the customer's ID with each request, then set an end-user budget for that ID.

Getting started

• Log in to the Mission Control portal and click on AI Gateway under the Inference section on the left-hand menu.
• Create a gateway and choose its models.
• Set its budget and rate limits, then add guardrails.
• Create an inference key for a person and pick which of the gateway's models it can use.
• Point your OpenAI or Anthropic SDK at https://llm.telnyx.com/v1 with that key.

Learn more in the AI Gateway docs or the API quickstart.