llms.txt
# Fabric Gateway
> Fabric Gateway is an AI-native gateway that routes, governs, and observes LLMs, MCP servers, agents, and internal APIs through a single endpoint deployed inside your own cloud.
Fabric Gateway is made by Postman ("Fabric from Postman"). Highlights: 50+ providers, 1,200+ models, LLM, MCP, A2A and API support, Flexible deployment options.
Every fabricgateway.ai page linked below has a Markdown version: https://fabricgateway.ai/index.md for the homepage, and the page's path without the trailing slash plus `.md` for every other page (for example https://fabricgateway.ai/pricing.md). Requests sent with `Accept: text/markdown` get the Markdown version too.
## Capabilities
- [Universal connectivity](https://fabricgateway.ai/product/universal-connectivity/): Connect every AI service. Route models, MCP servers, agents and internal APIs through one endpoint, with consistent authentication, governance and observability across every connection.
- [Policy-driven governance](https://fabricgateway.ai/product/policy-driven-governance/): Enforce policy on every call. Apply authentication, authorization, budgets and data controls centrally across teams, agents, models, tools and environments.
- [Credential brokering](https://fabricgateway.ai/product/credential-brokering/): Let agents act without exposing secrets. Agents use scoped Fabric tokens while the data plane retrieves and injects the correct upstream credential at request time.
- [AI-native workload support](https://fabricgateway.ai/product/llms/): Control AI-specific traffic. Fabric understands models, tokens, streaming responses and tool calls, enabling token-aware limits, resilient routing and provider fallbacks.
- [Full-stack observability](https://fabricgateway.ai/product/observability/): Trace complete agent workflows. Follow model, MCP, agent and API calls across one workflow, with identity, policy, latency, token and cost context.
- [Zero-trust agent security](https://fabricgateway.ai/product/zero-trust-agent-security/): Verify every agent action. Validate the initiating user and executing agent, restrict delegated permissions and deny requests when identity or policy cannot be verified.
## Product pages
- [Adaptive model routing](https://fabricgateway.ai/product/adaptive-model-routing/): Adaptive model routing in Fabric Gateway reads every prompt and matches it to an appropriate model in real time. Simple prompts stop paying for frontier models, your apps keep calling one endpoint, and output quality stays high.
- [Universal Connectivity](https://fabricgateway.ai/product/universal-connectivity/): Point your existing SDK at Fabric Gateway and reach 1,200+ LLM models, internal and external MCP servers, your own APIs, A2A agents, and gRPC services through a single control plane.
- [Policy-Driven Governance](https://fabricgateway.ai/product/policy-driven-governance/): Auth enforced at both edges of the gateway, inbound requests and outbound credential use, under one consistent policy layer.
- [Credential Brokering](https://fabricgateway.ai/product/credential-brokering/): Agents authenticate with a scoped gateway token. Fabric looks up and injects the real upstream credential inside your perimeter, never exposed to the agent.
- [Observability](https://fabricgateway.ai/product/observability/): Every call is instrumented automatically (LLM, MCP, agent-to-agent, or internal API) and pushed to the observability stack you already use.
- [Zero-Trust Agent Security](https://fabricgateway.ai/product/zero-trust-agent-security/): Dual identity validation, purpose-scoped token exchange, and fail-closed enforcement on every agent-to-agent request.
- [MCP Tool Search](https://fabricgateway.ai/product/mcp-tool-search/): Expose every registered MCP server behind a single endpoint, without loading every tool definition into the model's context window just to make one call.
## What it connects
- [LLMs across every major provider](https://fabricgateway.ai/product/llms/): 1,200+ models from 50+ providers through one OpenAI-compatible endpoint. Change the base URL; keep your SDK.
- [MCP servers, internal and external](https://fabricgateway.ai/product/mcp-servers/): Integrate with any external MCP server. Register your internal ones in minutes. Fabric governs both the same way.
- [Internal APIs, agent-ready](https://fabricgateway.ai/product/internal-apis/): Register APIs from Postman collections, OpenAPI / AsyncAPI, or your Private API Network. Agents reach them on the same control plane as LLMs and MCP.
- [Agents](https://fabricgateway.ai/product/agents/): Route A2A calls through the same gateway as everything else, with identity checked on both the agent and the human behind it.
## Zero-trust agent security
- Verify both identities: Check the initiating user and the executing agent on every request before anything runs.
- Limit every delegation: Issue purpose-scoped access per hop instead of forwarding the original user token.
- Authorize tool access: Allow only the tools and actions the agent is permitted to use.
- Deny by default: If identity or policy cannot be verified, Fabric denies the request.
## Deployment
[Enterprise Deployment](https://fabricgateway.ai/product/enterprise-deployment/): Run the data plane in your own cloud, let Postman operate it, or split the two. The endpoint your applications call stays the same.
- Self-hosted in your VPC: Run the data plane inside your own Kubernetes cluster or VM fleet.
- Postman-hosted: Let Postman operate it for you with regional data planes.
- Hybrid: Control plane managed by Postman; data plane stays in your cloud.
## Blog (https://fabricgateway.ai/blog/)
- [Adaptive Model Routing in Fabric Gateway: Cutting LLM Costs 50–80% on Most Workloads](https://fabricgateway.ai/blog/adaptive-model-routing-in-fabric-gateway/): How adaptive model routing in Fabric Gateway classifies each request once and routes it to a cheap, medium or strong model tier, cutting LLM costs 50–80% for a 1–3 point quality drop, with benchmark results for OpenAI and Anthropic.
## Pricing (https://fabricgateway.ai/pricing/)
### Free: $0/mo
For developers and teams shipping production traffic.
- Unlimited API, MCP, and AI requests
- Postman cloud hosted
- 50+ providers and 1,200+ models
- Automatic retries and fallbacks
- Audit logs (30 days) and traces (90 days)
- Usage metering and cost visibility
- Community and email support
### Enterprise: Custom
For organizations with governance, scale, and security requirements. Everything in Free, plus:
- Hybrid deployment — data plane in your VPC
- Granular RBAC, SSO, and admin controls
- Custom request limits and retention
- Custom platform fees and usage terms
- Dedicated solution engineering
- Priority support and SLA options
## FAQ
### What is included in the Free plan?
Free runs entirely on Postman cloud with no platform fees. You get unlimited API, MCP, and AI requests, core observability, and usage metering. Some AI and MCP capabilities ship in open beta while we expand the surface.
### Do I pay for LLM tokens through Fabric?
No Fabric meter applies to LLM provider traffic on Free. You pay your model provider at their list price. Fabric gives you routing, policy, and observability on that traffic without an additional gateway fee.
### What is granular access control?
Enterprise lets you define fine-grained permissions across resources, routes, credentials, and admin actions. Free includes standard team collaboration but not RBAC-level permission scoping or admin controls.
### What is the difference between Free and Enterprise hosting?
Free runs both the control plane and data plane in Postman cloud — fastest path to production. Enterprise keeps the control plane on Postman while you run data plane nodes in your own cloud for data residency and network isolation.
### Can I change plans later?
Yes. Most teams start on Free and move to Enterprise when security review requires a customer-hosted data plane, SSO, granular RBAC, longer retention, or committed usage. Book a demo to discuss migration.
## Links
- [Homepage](https://fabricgateway.ai/)
- [Pricing](https://fabricgateway.ai/pricing/)
- [Book a demo](https://fabricgateway.ai/book-demo/)
- [Documentation](https://docs.fabricgateway.ai)
- [Documentation index for AI assistants](https://docs.fabricgateway.ai/llms.txt): an index of the docs, with a markdown version of every page
- [Quickstart](https://docs.fabricgateway.ai/get-started/quickstart)
- [Docs MCP server](https://docs.fabricgateway.ai/_mcp/server): connect Claude Code, Cursor and other AI clients to the docs
- [Sitemap](https://fabricgateway.ai/sitemap.xml)
- [Sign up and log in](https://app.fabricgateway.ai)
