Product · Beta

AI Gateway
The Hub for Enterprise AI Procurement & Governance

One endpoint for all AI capabilities—unified procurement, governance, security control, and billing.One API connects 6 LLM providers—OpenAI, Anthropic, Azure, Bedrock, Cohere, Vertex—with BYOK support and automatic fallback.

Or leave your contact info

Core Capabilities

Unified access and management of multiple models, with secure integration and governance

AI capabilities behind a single endpointChat LLMEmbeddingsRerankerDocument ParserSpeech STT/TTSImage GenerationSearch & Data Services

OpenAI-Compatible API

OpenAI-compatible API—just switch the base URL. A single entry point and single endpoint for all AI capabilities, integrating quickly with existing applications.

6 LLM Providers, One API

One API connects OpenAI, Anthropic, Azure, Bedrock, Cohere, and Vertex, with 10+ upstream providers and counting.

Model Routing & Auto Fallback

Flexible routing across models and providers, automatically dispatching to the best path—with automatic fallback for high availability.

Security Governance & Access Control

Data filtering, risk control, model allowlists, rate limiting, and permission management—compliance-ready end to end.

Usage Logs & Billing Integration

Unified billing with a single invoice and shared wallet. Usage monitoring and billing management in one place—AI costs stay predictable and controllable.

Complete Audit Trail

Per-request records and reconciliation reports, fully auditable for compliance—every AI call leaves a complete audit trail.

Deployment Options

Three scenarios, three pricing models—choose what fits your enterprise

On-Premises Deployment

On-prem customers procure external AI resources through the Gateway

  • Only one outbound connection to MaiAgent Gateway required
  • On-prem masking and security hooks—data is masked before leaving
  • Annual prepaid resource pack, shared across the organization
  • 80% / 100% usage alerts with grace buffer—no interruption
  • Per-request auditing and reports, read-only on-prem dashboard

Pricing

License fee + prepaid usage pack (annual prepayment)

SaaS Edition

LLM Gateway as a cloud service by MaiAgent

  • OpenAI-compatible API—just switch the base URL
  • Gateway shares the same wallet as chat assistants, one unified bill
  • Multi-model routing with platform-wide models enabled by default
  • Full protection: rate limits, quotas, and concurrency control
  • Per-organization configurable model allowlist

Pricing

Usage-based billing (unified invoice)

BYOK (Bring Your Own Key)

Bring your own model keys; MaiAgent charges a data processing fee

  • Customers bring their own model keys (BYOK)
  • Billed by data volume processed (per 1K tokens)
  • Model-agnostic—you pay for data processing, not models
  • Combine with security hooks for stronger governance and auditing

Pricing

Data processing fee (usage-based)

Why an AI Gateway?

The same model can be served by different providers and regions. Latency mostly comes from the network and provider load—not the model itself—so switching providers or regions really works.

1

Network Distance

Determined by region

Round-trip time to the region hosting the model—the closer, the faster. Choosing a region near your users significantly reduces network latency.

2

Provider Load

Determined by provider

Queuing, capacity, and throttling at the provider affect wait times. For the same model, provider A being congested while provider B runs smoothly is common.

3

Model Computation

Essentially fixed

For the same model, computation time is essentially fixed. Switching providers or regions does not change it—the bottleneck usually lies in the first two.

AI Gateway monitors latency and availability across providers and regions in real time, automatically selecting the best path for a stable experience.

Get Early Access to AI Gateway

AI Gateway is currently in beta. Contact us to learn how to govern your enterprise AI resources through a single entry point.

LINE