Guardrail Layer — The LLM API Gateway With Built-in Observability & Guardrails
Features How It Works Providers Docs
Sign In Get API Key →
In BETA

Control every AI request
your app makes.

The LLM API gateway that logs requests, enforces spend limits, replays calls, and routes across providers — all through a single API.

Get Started Free → 📄 View Docs ⚡ Try Demo
guardrail layer — request logs
Timestamp Model Latency Cost Status Action
14:32:08.412 gpt-4o 1.24s $0.0083 ● 200
14:32:06.991 claude-sonnet 0.89s $0.0041 ● 200
14:32:04.223 gpt-4o — — ● 429
14:32:01.887 gemini-pro 0.67s $0.0012 ● 200
14:31:59.102 openrouter 0.54s $0.0019 ● 200
// the problem

AI apps are hard to control.
You need an LLM API gateway.

You're shipping fast but flying blind. One bad loop can burn your budget. One silent failure can tank your product.

💸

Costs spike unexpectedly

A recursive loop at 3am burns through your monthly budget in minutes. No alerts, no limits.

🔇

Requests fail silently

Rate limits, timeouts, malformed responses — your users see broken features while you see nothing.

🐛

Debugging is painful

Reproducing an LLM bug means guessing at the prompt, context, and parameters. Every time.

🔀

Switching providers is messy

Each provider has different APIs, auth, and response formats. Migrating is a rewrite.

guardrail.layer — the LLM API gateway that fixes this ↓
// core features

One LLM API gateway.
Everything you need to ship AI.

Four capabilities that give you complete visibility and control over every LLM call.

01 — Observability

Real-time
Request Logs

Every request that flows through the LLM API gateway is logged with full context. See exactly what's happening across your entire AI stack.

  • ✓ Model, provider, and endpoint
  • ✓ Cost per request, cumulative spend
  • ✓ Latency and token usage
  • ✓ Status codes and error details
live request stream
✓ 14:32:08 gpt-4o 1.24s $0.008 432 tokens
✓ 14:32:06 claude-sonnet 0.89s $0.004 287 tokens
✗ 14:32:04 gpt-4o — 429 rate limited
✓ 14:32:01 gemini-pro 0.67s $0.001 156 tokens
🔥 02 — Debugging

Instant
Request Replay

Click any logged request and replay it instantly — same prompt, same parameters, same context. Change one variable and see what happens. This is the feature devs dream about.

  • ✓ One-click replay from logs
  • ✓ Modify and re-send with diff view
  • ✓ Compare original vs replay responses
  • ✓ Reproduce bugs in seconds, not hours
replay request #4821
Replaying request to gpt-4o ● Ready
"model": "gpt-4o",
- "temperature": 0.9,
+ "temperature": 0.3,
"messages": [
{"role": "user",
- "content": "Summarize the doc"}
+ "content": "Summarize the doc in 3 bullets"}
]
03 — Cost Control

Automatic
Spend Protection

Set dollar limits at the project, key, or model level. When you're close, get warned. When you hit it, requests are blocked or downgraded automatically.

  • ✓ Per-key and per-model spend limits
  • ✓ Auto-downgrade to cheaper models
  • ✓ Real-time spend tracking dashboard
  • ✓ Email/webhook alerts at thresholds
spend protection — project: acme-chatbot
$36.14 of $50.00 limit
⚠ 72% of monthly limit Alert at $40 · Block at $50
12.4K
Requests
0.89s
Avg Latency
$0.003
Avg Cost
04 — Routing

Multi-Provider
Routing

OpenAI, Anthropic, Google, OpenRouter — one unified LLM API gateway. Switch providers with a config change, not a code rewrite. Automatic fallback when a provider is down.

  • ✓ One API for all LLM providers
  • ✓ Automatic failover on errors
  • ✓ Route by cost, latency, or model
  • ✓ Drop-in replacement for OpenAI SDK
provider routing
APP
your code
GRL
guardrail
OAI
ANT
GEM
ORT
Request → Log → Route → Deliver → Log Response
// integration

Up and running in 60 seconds.

Connect your app to the LLM API gateway in three steps. No SDK. No config files. Just point your requests through us.

1

Add your API key

Generate a Guardrail Layer key in the dashboard. Link your provider keys.

2

Send requests through us

Change your base URL to Guardrail Layer. That's the only code change.

3

See everything instantly

Every request appears in your dashboard. Set limits, replay calls, explore data.

bash
curl https://guardraillayer.com/api/v1/chat/completions \ -H "Authorization: Bearer grl_live_38867...2dd47e1" \ -H "Content-Type: application/json" \ -d '{ "model": "claude-sonnet-4-20250514", "messages": [{"role":"user","content":"Hello"}] }'
🔌

Drop-in Replacement

Compatible with the OpenAI SDK. Change one line and your app runs through the LLM API gateway.

🔓

No Vendor Lock-in

Your keys, your providers, your data. Remove Guardrail Layer anytime with zero migration cost.

⚡

<5ms Overhead

Transparent proxy adds minimal latency. Your users won't notice. Your logs will thank you.

// supported providers

An LLM API gateway for every major provider.

// get started

Start controlling your AI today.

Free tier. No credit card. Full observability in under a minute.

Scroll to Top