OpenAI-compatible · Groq · Cerebras · HuggingFace

Route LLMs for free.
Zero cost, zero friction.

One API key. Auto-failover across the fastest free-tier LLM providers. Drop-in OpenAI replacement — change one line of code.

01

Get your free key

Enter your email below. Your zc-key arrives instantly.

zc-xxxxxxxxxxxxxxxxxxxxxxxx
02

Register your providers

Send your Groq, Cerebras, or HuggingFace keys once.

curl -X POST https://zerocost-router.dragonrondo.workers.dev/v1/keys/providers \
  -H "Authorization: Bearer zc-your-key" \
  -H "Content-Type: application/json" \
  -d '{"provider_id":"groq","api_key":"gsk_..."}'
03

Make your first API call

Point your client at zerocost. Done.

from openai import OpenAI

client = OpenAI(
  base_url="https://zerocost-router.dragonrondo.workers.dev/v1",
  api_key="zc-your-key",
)

response = client.chat.completions.create(
  model="llama-3.3-70b-versatile",
  messages=[{"role": "user", "content": "Hello!"}],
)

Built for developers who don't want to pay for LLM APIs

Multiple free LLM providers. One endpoint. Automatic failover between them — so rate limits become someone else's problem.

Auto-Failover

When one provider hits a rate limit, zerocost silently switches to the next. Your requests keep flowing — no retries, no error handling needed.

OpenAI Compatible

Drop-in replacement for the OpenAI client. Change one line — the base URL — and every model call works across all providers.

Bring Your Own Keys

Connect your own Groq, Cerebras, and HuggingFace API keys. Your credentials are encrypted with AES-256-GCM and never logged.

Edge-Native Speed

Deployed on Cloudflare Workers — sub-50ms routing overhead from 300+ locations worldwide. No cold starts, no server maintenance.

Live provider status

live

Real-time quota data collected every hour from each provider's API.

Fetching live data…

Get your free API key

Enter your email. Your zc-key is generated instantly — no credit card, no waitlist.

No spam. Used only to identify your key.

Use with AI agents

Copy the prompt below and paste it into your AI agent. It will set up zerocost automatically — no manual steps needed.

setup prompt
Set up zerocost-router for me.

## What is zerocost-router?
A proxy API that provides an OpenAI-compatible endpoint backed by free-tier LLM providers (Groq, Cerebras, HuggingFace).
Endpoint base URL: https://zerocost-router.dragonrondo.workers.dev/v1

## Setup steps

### Step 1: Issue a zc-key
curl -s -X POST https://zerocost-router.dragonrondo.workers.dev/v1/keys
→ Save the "key" value from the response (e.g. zc-xxxxxxxx...)

### Step 2: Get a free Groq API key
Sign up at https://console.groq.com and generate an API key.

### Step 3: Register the provider key
curl -X POST https://zerocost-router.dragonrondo.workers.dev/v1/keys/providers \
  -H "Authorization: Bearer <zc-key>" \
  -H "Content-Type: application/json" \
  -d '{"provider_id":"groq","api_key":"<groq-api-key>","priority":0}'

### Step 4: Verify it works
curl -X POST https://zerocost-router.dragonrondo.workers.dev/v1/chat/completions \
  -H "Authorization: Bearer <zc-key>" \
  -H "Content-Type: application/json" \
  -d '{"model":"gpt-4o-mini","messages":[{"role":"user","content":"Hello"}],"max_tokens":50}'

### Step 5: Use with OpenAI SDK
base_url = "https://zerocost-router.dragonrondo.workers.dev/v1"
api_key  = "<zc-key>"
Use these settings in place of the standard OpenAI client configuration.

Once done, let me know the zc-key you received.

Works with any OpenAI-compatible AI agent or coding assistant.

Check your usage

Enter your zc-key to see how many requests you've used this month.

One key. Multiple providers. More headroom.

Each provider has its own rate limit. zerocost distributes your requests across all of them — so you get the sum, not just one.

Single provider

Provider A100%
Hits the rate limit on its own

zerocost (3 providers)

Provider A33%
Provider B33%
Provider C33%
Distributes load — up to 3× capacity
up to 3×

Actual limits vary by provider and model. Effective throughput depends on the number of providers you register.

Pricing

Start free. Scale when you need more.

Free

¥0forever

For personal projects and exploration.

  • 1,000 requests / month
  • 3 providers (Groq, Cerebras, HF)
  • Auto-failover
  • OpenAI-compatible API
Get started free

Basic

¥500/ month

For developers building apps.

  • 5,000 requests / month
  • 5 providers
  • Auto-failover
  • OpenAI-compatible API
  • Priority routing
Get Basic
Popular

Pro

¥1,500/ month

For teams and production workloads.

  • Unlimited requests
  • All providers
  • Smart routing (quota-aware)
  • OpenAI-compatible API
  • SLA-backed uptime
  • Usage analytics
Get Pro

FAQ

Which models are available?

Any model your registered providers support — including Llama 3.3 70B via Groq, Llama 3.1 70B via Cerebras, and a range of HuggingFace Inference models. zerocost doesn't restrict which model ID you pass.

Are my API keys safe?

Yes. Provider keys are encrypted with AES-256-GCM before storage. zerocost never logs request bodies or response content. Your keys cannot be retrieved in plaintext after registration.

What happens when all providers hit rate limits?

zerocost returns a clear error rather than silently failing. Rate limit flags reset automatically within 60 seconds, so subsequent requests will route normally.

Do I need a credit card?

No. The Free tier is completely free, no card required. Basic and Pro plans are billed securely via Stripe.

Start routing for free

Get your API key in 10 seconds. No card, no waitlist.

Get your free key →