Built for enterprise teams

AI Gateway for Enterprise Teams

Secure, compliant, and scalable AI infrastructure. Deploy 600+ models with enterprise-grade governance, security, and observability.

View documentation

Or email us at sales@requesty.ai

600+

AI models

225B+

Tokens per day

99.99%

Uptime SLA

131

EU ZDR endpoints

Trusted by leading companies

00The control plane

Every control, on every request

Five checks between your SDK call and the provider, and a record after it.

Stage 1 of 8, CallerSDKOne base URL, unchanged application coden/a
Stage 2 of 8, IdentityAuthSSO via Okta, key mapped to a team0.4ms
Stage 3 of 8, PolicyBudgetsupport, $1,800/mo cap, 67% used0.3ms
Stage 4 of 8, PolicyAllowlist3 of 600+ models cleared for this team0.2ms
Stage 5 of 8, GuardrailRedact3 of 8 PII types matched and scrubbed1.8ms
Stage 6 of 8, RoutingRegionPinned eu-central-1, Frankfurt0.6ms
Stage 7 of 8, UpstreamProvideropus-4.6, warm prefix at cache-read raten/a
Stage 8 of 8, RecordAuditActor, action, origin, streamed to your SIEMasync

Five checks, 3.3ms in front of a call that takes seconds.Measured at the Frankfurt gateway, p50

01GovernGovernance

Enterprise governance

Manage your entire AI infrastructure with precision. Set policies, control access, and maintain compliance across every team.

Permissions by role4 ROLES

What each role may do
Role	API	Budgets	Billing	Policies
Owner	Yes	Yes	Yes	Yes
Admin	Yes	Yes	No	Yes
Member	Yes	No	No	No
Viewer	No	No	No	No

Four roles, and the matrix is the whole definition: there is no fifth column hidden in a plan tier.

Budgets by scopeCASCADING

Allowlistopus-4.6gpt-5.4gemini-3.6-flash3 of 600+

Spending limits by scope
Scope	Limit	Used
Organisation acme-inc	$8,000/mo	61%
Team platform	$3,200/mo	54%
Team support	$1,800/mo	67%
Team research	$900/mo	92% near cap
Group contractors	$400/mo	18%

A limit set on the organisation caps every team under it. research is inside its own cap and inside the organisation's.

Audit logEXPORTABLE

Recent audit events
Time	Action	Origin
14:22	j.okafor raised the research budget to $900/mo	SSO
14:09	ci-eval-runner key rotated	API
13:54	p.nowak pinned acme-prod to eu-central-1	SSO
13:31	d.mensah added gpt-5.4 to the support allowlist	SSO
12:47	a.silva revoked 3 contractor keys	SCIM

Every action, actor, timestamp and origin. Streamed to your SIEM or exported as JSON.

02SecureSecurity

Security and compliance

Built with enterprise security at the core. Your data stays private, protected, and under your control at all times.

Compliance register4 CONTROLS

SOC 2 Type IIIn progressObservation period under way with an independent auditor

GDPRCompliantFull EU data protection regulation compliance

DPAOn requestArticle 28 processor agreement, signed at any spend

ISO 27001In progressTarget date published on the trust page

SOC 2 Type II is an observation period, not a certificate. The target date is on the trust page.

Data handling8 ENTITY TYPES

Your application sends

Customer sarah.chen@northwind.co called about invoice 8841. Callback +44 7700 900412, card ending 4242.

Scrubbed at the gateway, 3 of 8 types matched

The model receives

Customer [EMAIL] called about invoice 8841. Callback [PHONE], card ending [CARD].

Encryption and privacyEnd-to-end encryption in transit and at rest. Zero data retention policy. We never train on your data or store prompts beyond processing.

Data residencyChoose where your data lives. EU requests stay in Frankfurt, US in Virginia. A call that would leave the pinned region is refused, not forwarded.

Automatic detection and redaction of personal data before it reaches any model. Detection runs in your region: email, phone, credit card, national ID, IBAN, IP address, postal address and date of birth.

At scale

The gateway is the boring part

Three numbers decide whether an AI layer stops being a topic.

03RouteInfrastructure

Intelligent infrastructure

Enterprise-grade routing, caching, and reliability. Built for production workloads at any scale.

Failover, last event14MS TOTAL

StepTook

Endpoint degraded2ms

Health check6ms

Pick next provider3ms

Switch traffic3ms

Total14ms

The retry happens inside the gateway, so the caller sees one request. Warm prefixes bill at the provider's cache-read rate.

Region routingONE ENDPOINT

Gateway regions and the policy applied to each
Region	Serves	Policy
eu-central-1pinned	EU, Frankfurt	EU models only
us-east-1	US, Virginia	nearest healthy
ap-southeast-1	APAC, Singapore	nearest healthy

Route to the nearest region automatically, or pin one. The routing policy on the key decides, not the caller.

04SeeObservability

Enterprise observability

Complete visibility into your AI infrastructure at scale. Monitor costs, performance, and usage across all teams and providers.

Enterprise cost and performanceLAST 30D

30d spend

$12,470

Requests

1.94M

p95 latency

950ms

Success rate

99.85%

Cost / day

$415

p95 latency

950ms

Requests / hr

2.7K peak

Errors, 24h

Track spending by model, team, user and project in real time, and set alerts before budgets are exceeded. Alerts fire on the p95, not the mean.

By agent92.3K CALLS, 30D

support-triageHealthy42.1K calls$182

p50 165msp95 500ms

code-reviewerHealthy18.8K calls$126

p50 545msp95 1.4s

data-extractorHealthy31.4K calls$259

p50 285msp95 835ms

Track latency, cost and success rates per agent. Identify bottlenecks, optimise routing, and hold an SLA for every agent in the fleet.

Usage by teamLAST 30D

platform41%795K

support26%504K

research19%368K

growth14%271K

Understand which models, teams and users drive consumption, so model selection and capacity planning start from data rather than from an argument.

One line to integrate

OpenAI SDK compatible

Switch in seconds. Change one line of code and you're running on Requesty with full enterprise governance.

Works with existing code
Existing SDK calls unchanged
Instant governance layer

integration.py

# Before: OpenAI
client = OpenAI(
    api_key="sk-..."
)

# After: Requesty
client = OpenAI(
    api_key="req-...",
    base_url="https://router.requesty.ai/v1",
)

# That's it. Full enterprise governance applied.

EU customers swap the host for router.eu.requesty.ai. Same API, same key, and the request terminates in Frankfurt.

Simple process

How it works

Get started with enterprise-grade AI infrastructure in three straightforward steps.

Step 01

Talk to our team

Schedule a call to discuss your requirements, compliance needs, and integration points. We’ll design a solution tailored to your organization.

Step 02

Custom onboarding

We’ll set up your workspace with your policies, connect your IdP, configure team permissions, and migrate your existing infrastructure.

Step 03

Go live

Deploy with one line of code. We provide hands-on support during migration, optimization, and scaling to ensure a smooth transition.

Enterprise FAQ

Common questions from enterprise teams evaluating Requesty.

Does Requesty support SSO?

Yes. Requesty supports SSO via Okta, Azure AD, and Google Workspace on enterprise plans, with full SAML and OIDC support.

Is there a free trial?

Yes. You can start on the pay-as-you-go plan with no commitment, then upgrade to enterprise when you need SSO, RBAC, dedicated support, or custom SLAs.

Can Requesty be self-hosted?

Not today. Requesty is a fully managed cloud platform with EU data residency available in Frankfurt. Self-hosting is not offered at this time.

How does RBAC work?

Role-based access is fully managed on the platform. Admins see all activity, spend, and keys across the organization. Users only see their own keys, usage, and logs. Groups and team-level budgets let you delegate further.

What models can our team use?

All 600+ models by default. Enterprise plans let admins restrict access to an approved list of models and providers, so users can only call models your organization has vetted.

How is enterprise pricing calculated?

Enterprise plans are priced on request. Contact our team to discuss your volume, required features (SSO, RBAC, guardrails, audit logs), and support needs.

Provenance. Model count (600+), tokens per day (225B+), EU zero-retention endpoints (131) and the 14ms failover are platform measurements. Gateway overhead of 3.3ms is the sum of the five checks shown above, measured at the Frankfurt gateway at p50. Uptime SLA of 99.99% is a contractual commitment, not an observed result. SOC 2 Type II is an observation period under way, not a certification. Customer logos are the marks of Requesty customers as published on /customers.

Ready to deploy AI at scale?

Get in touch with our team to discuss your requirements and see how Requesty can transform your organization's AI infrastructure.

View documentation

Or email sales@requesty.ai.