Fabric Gateway

What it does

Fabric Gateway takes in requests from applications, SDKs, AI agents, MCP clients, and HTTP services, and routes them to LLM providers, MCP servers, agents, and internal APIs. Its control plane manages policy, identity, and configuration, while a data plane routes live traffic within the chosen deployment environment. Outputs include routed model responses, retries and fallbacks, and audit logs, traces, and cost or usage data.

Teams use it to connect many AI services through one endpoint rather than integrating each provider separately. The page describes changing an SDK's base URL to point at Fabric, with credentials brokered so agents receive scoped Fabric tokens instead of upstream keys. It also highlights health-aware routing, circuit breakers, token-aware rate limits, per-team budgets, and tracing of model, MCP, and API calls within a single workflow.

It suits platform, security, and engineering teams that need central governance across many models, tools, and agents. The Free plan has no platform fee and includes API, MCP, and AI requests, with audit logs kept 30 days and traces 90 days, but some AI and MCP capabilities are in open beta. Enterprise adds hybrid deployment, SSO, granular RBAC, and custom terms through a demo request, and RBAC is not included on Free.

What it can do

• Routes LLMs across 50+ providers and 1,200+ models through one OpenAI-compatible endpoint.
• Connects MCP servers, internal REST and gRPC APIs, and agents through a single gateway with consistent authentication and observability.
• Brokers credentials so applications and agents never hold upstream provider keys.
• Applies health-aware routing, retries, circuit breakers, and automatic fallbacks across providers.
• Enforces token-aware rate limits, per-team budgets, and centralized policy across teams and workloads.
• Provides end-to-end traces, audit logs, and usage and cost visibility across models and services.

Works with

• OpenAI-compatible SDKs via base URL change
• Ollama Cloud
• Hugging Face
• Snowflake Cortex
• SAP AI Core
• Weights & Biases Inference

Before you choose

• Some AI and MCP capabilities ship in open beta, and Free does not include RBAC-level scoping or admin controls.
• Free retains audit logs for 30 days and traces for 90 days; longer retention and SSO require Enterprise.
• LLM token costs are billed by the model provider at list price; Fabric adds no gateway fee on Free.

Listed under

• Coding
• MCP and agent tools

Get started

Start from the sign-up link; docs are available for setup.

1. Create an account via the sign-up link on the homepage.
2. Point a supported SDK at the Fabric base URL, https://api.fabricgateway.ai/v1, changing only the base URL.
3. Register your LLMs, MCP servers, agents, or internal APIs in the gateway and apply policies, then review the docs at docs.fabricgateway.ai.

Free has no platform fee and no Fabric meter on LLM provider traffic; you pay providers directly. Enterprise terms are custom, including platform fees and usage terms.

Source: fabricgateway.ai/pricing, checked 30 Oct 2026. Prices can change; check before you buy.

Sources for this overview

Product pages, documentation, and pricing checked 30 Oct 2026. Each source supports part of the overview; availability and prices can change.

• Fabric Gateway — AI gateway for LLMs, MCP servers, and agents ↗fabricgateway.ai
• Pricing — Fabric Gateway ↗fabricgateway.ai

Company website

fabricgateway.ai: product, pricing and policy pages

Social accounts

Accounts the company links, and its posts on X

Human review

Checked by a person

knowledge_base

manual_review

Last checked 30 Oct 2026. Every field keeps its source and date; unknown means we couldn’t confirm it.