One API for every LLM.
Secured at the edge.

Route 26 models across 7 providers. Block prompt injections and data leaks. Audit every AI request. Change one line of code.

26 Models, 9 Tiers

<1ms Security Scan

45+ Integrations

See How It Works Create Account

The problem with AI in production

Most teams bolt LLM calls into their apps with no routing, no security, and no visibility.

💰

You're overpaying

No smart routing means every prompt hits GPT-4. 40% of requests could go to lower-cost models with zero quality loss.

👁

You're blind

No audit trail for employee AI usage. No idea what data is being shared with third-party models.

🔒

You're exposed

No injection detection. No DLP. One prompt away from a data breach that leaks your system prompt or API keys.

How ModelRouter works

A single edge layer between your app and every LLM provider.

Your App

→

ModelRouter

→

Secure Route Cache Log

→

7 Providers

Step 1 — Secure

Injection detection & DLP

Every prompt is scanned for 30+ injection patterns and sensitive data (SSNs, API keys, source code) before it leaves the edge. Malicious requests are blocked in <1ms.

Step 2 — Route

Smart model selection

Prompts are classified into 9 tiers by complexity and content type. Specialized tasks route to specialized models: OCR to Mistral OCR (not GPT-4o), image generation to FLUX.1 via Together AI (not just DALL-E), code to DeepSeek Coder, reasoning to Claude Opus 4 -- all automatically based on your cost, speed, and quality preferences.

Step 3 — Cache

2-tier edge caching

Identical or semantically similar prompts are served from Fastly's edge cache. Average 30% cache hit rate eliminates repeat LLM spend entirely.

Step 4 — Log

Full audit trail

Every request is logged with provider, model, cost, latency, security verdict, and user context. Export to CSV anytime for compliance.

Try it right now

Click a prompt to see the full routing pipeline in action.

Live Pipeline Demo

Receive prompt at edge

—

Injection scan (30+ patterns)

—

DLP check (PII, secrets)

—

Classify complexity tier

—

Route to optimal provider

—

Return response + log

—

Built for security teams

Every AI request passes through three layers of protection before reaching any provider.

🛡

Injection Detection

CPU-only pattern matching catches jailbreaks, instruction overrides, and system prompt extraction attempts.

<1ms · 30+ patterns · Zero false positives on normal prompts

📛

DLP Rules

Block prompts containing SSNs, credit card numbers, API keys, or source code before they leave your network.

Configurable rules · Regex + keyword · Per-org policies

📋

Audit Logging

Every request logged with user, prompt hash, provider, model, cost, latency, and security verdict.

Per-request · CSV export · Compliance-ready

One line of code

Drop-in OpenAI SDK compatible. Change your base URL and you're done.

# Before
from openai import OpenAI
client = OpenAI(api_key="sk-...")

# After — that's it, one line
client = OpenAI(
    base_url="https://your-service.edgecompute.app/v1",
    api_key="YOUR_MODELROUTER_KEY"
)

// Before
import OpenAI from "openai";
const client = new OpenAI({ apiKey: "sk-..." });

// After — that's it, one line
const client = new OpenAI({
    baseURL: "https://your-service.edgecompute.app/v1",
    apiKey: "YOUR_MODELROUTER_KEY"
});

# Before
curl https://api.openai.com/v1/chat/completions \
  -H "Authorization: Bearer sk-..." \
  -d '{"model":"gpt-4o","messages":[...]}'

# After — change the URL, that's it
curl https://your-service.edgecompute.app/v1/chat/completions \
  -H "Authorization: Bearer YOUR_MODELROUTER_KEY" \
  -d '{"model":"auto","messages":[...]}'

Cut your LLM spend by 60%

Smart routing + caching + included-tier models = dramatic cost reduction.

30%

Cache hit rate eliminates repeat LLM calls entirely

40%

Of requests route to included-tier models with zero quality loss

Markup on provider costs — bring your own keys

See pricing calculator →

How we compare

ModelRouter is the only gateway that combines edge deployment, security, and cost optimization.

Feature	ModelRouter	OpenRouter	Portkey	LiteLLM
Edge deployment	Yes (Fastly)	No	No	No
Injection detection	Yes (30+ patterns)	No	No	No
DLP / blocking rules	Yes	No	Basic	No
Audit logging	Yes (per-request)	No	Yes	No
Edge caching	Yes (2-tier)	No	Yes	No
Vendor-independent (BYOK)	Yes	5.5% fee	Per-request	Yes (OSS)
Multimodal routing	Yes (auto-detect)	No	No	No
Included-tier routing	Yes (auto)	No	No	No
Self-hostable	Yes	No	No	Yes
Cold start	<1ms (WASM)	N/A	N/A	N/A

Stop AI Sprawl

Every AI provider is a separate vendor, a separate data processor, a separate discovery target. ModelRouter consolidates them behind one gateway with one audit trail.

Read: AI Sprawl & Legal Discovery →

Simple, transparent pricing

Start small. Scale when you're ready. No markup on provider costs.

Starter

$0/mo

1,000 requests/day
All providers (included tiers)
Injection detection
Basic audit log (7 days)
Community support

Ready to secure your AI?

Create Account

Talk to the Founder →

One API for every LLM.Secured at the edge.