ZERO COST. MAX ROUTE.
The $0/mo Multi-Cloud AI Gateway for Solo Founders & Startups.
Pool 100% free cloud LLM quotas across Groq, SambaNova, Mistral, Google Gemini, NVIDIA NIM, Cloudflare, and Cohere into one unstoppable, zero-downtime OpenAI-compatible API.
As solo founders, indie hackers, and small business owners, we want to add intelligent AI chatbots, customer support assistants, and content generators to our websites and portfolio apps.
However, paying \ to +/month for commercial LLM APIs before finding product-market fit can be prohibitive.
Meanwhile, top cloud AI infrastructure companies (Groq, SambaNova, Mistral, Google Gemini, OpenRouter, NVIDIA NIM, Cloudflare, Cohere) offer generous 100% FREE developer tiers. The catch? Any single free tier has rate limits (RPM / RPD) or occasional congestion.
ZeroRoute aggregates all 8 free cloud tiers into one unified, resilient, high-speed OpenAI-compatible gateway.
- If Groq is busy, it automatically fails over to SambaNova in milliseconds.
- If SambaNova hits a rate limit, it falls back to Mistral, Gemini, OpenRouter, or NVIDIA NIM.
- Exact queries are served from in-memory RAM cache at 0ms latency for .00.
- Result: You get millions of monthly tokens and near 100% uptime for .00/month!
| Provider | Default Verified Model | Speed / Latency | Key Strengths |
|---|---|---|---|
| ⚡ Groq | qwen/qwen3.6-27b / openai/gpt-oss-120b | ~100ms | Ultra-fast LPU inference, instant streaming |
| 🔥 SambaNova | MiniMax-M2.7 / gemma-4-31B-it | ~360ms | 417+ tokens/sec throughput, SN40L chips |
| 🌪️ Mistral AI | mistral-medium-latest / codestral-latest | ~390ms | European privacy, specialized coding & reasoning |
| 💎 Google Gemini | gemini-3.5-flash-lite / gemini-flash-lite-latest | ~710ms | 1M+ token context window, Google multimodal |
| 🌐 OpenRouter | |||
| vidia/nemotron-3.5-lightning:free | ~1500ms | 100% free open-source models aggregator | |
| 🚀 NVIDIA NIM | |||
| vidia/nemotron-3.5-lightning-30b-a3b | ~260ms | Accelerated enterprise DGX cloud infrastructure | |
| ☁️ Cloudflare Workers AI | @cf/meta/llama-3.1-8b-instruct | ~1100ms | Global edge network with 300+ datacenters |
| ⚛️ Cohere | command-r-plus-08-2024 | ~1000ms | Enterprise conversational reasoning |
- 🔄 Zero-Downtime Provider Failover: Automatic multi-hop routing if any upstream provider encounters a 429, 500, or Timeout.
- ⚡ Parallel Concurrency Benchmark Suite: Races all 8 providers in parallel (~1.5s total) with an animated progress bar, winner podium, and 1-Click Optimal Order Sorting.
- 🔄 Self-Healing Background Heartbeat: Automatically probes cooling providers every 30s and restores them to active traffic the instant their rate limit clears.
- ⚡ Real-Time SSE Streaming ( ext/event-stream): Smooth token-by-token streaming for web chatbots and terminal clients.
- 💾 In-Memory SHA-256 Response Cache: Serves identical requests instantly at 0ms latency with live dollar savings estimation.
- 🛡️ AES-256-GCM Encrypted Key Storage: API keys are encrypted at rest on disk with PBKDF2 key derivation and never exposed over client network payloads.
- 🎛️ Obsidian & Crimson Brand Dashboard: Interactive model catalog modal, raw error inspector, request waterfall trace, and live playground with custom system persona prompts.
- 📦 Zero Runtime NPM Dependencies: Built purely with native Node.js HTTP and Fetch APIs.
Deploy your own private AI Gateway to Vercel in 30 seconds:
- Click the Deploy with Vercel button above.
- Paste at least one free API key (e.g. GROQ_API_KEY, MISTRAL_API_KEY, or GEMINI_API_KEY).
- Your gateway and live dashboard are instantly live globally on https://your-project.vercel.app!
\\�ash git clone https://github.com/amjadlle/ZeroRoute.git cd ZeroRoute cp .env.example .env npm install \\
Open .env and add any free API keys you have: \\env
GROQ_API_KEY=gsk_... SAMBANOVA_API_KEY=... MISTRAL_API_KEY=... GEMINI_API_KEY=AIzaSy... \\
\\�ash
npm run dev
\
Visit http://localhost:8787 to open your live dashboard!
Run with Docker Compose in 1 command: \\�ash docker compose up -d \\
Drop this single line of HTML anywhere in your website (index.html, Next.js, React, Vue, WordPress, Webflow, or Shopify) to add an AI floating chatbot bubble powered by your free cloud LLM tiers:
<script
src="https://your-zeroroute.vercel.app/widget.js"
data-title="Amjad AI"
data-persona="You are Amjad's portfolio assistant. Answer questions politely and concisely."
data-greeting="Hi there! 👋 How can I help you today?"
data-color="#ef4444"
defer>
</script>You can attach your company FAQs, bio, documentation, or pricing by passing inline text OR a remote .md / .txt / .json file URL:
<script
src="https://your-zeroroute.vercel.app/widget.js"
data-title="Acme Support AI"
data-persona-url="https://your-website.com/persona.txt"
data-knowledge-url="https://raw.githubusercontent.com/username/repo/main/faq.md"
data-color="#ef4444"
defer>
</script>| Attribute | Description | Example |
|---|---|---|
data-title |
Header title of the chatbot window | Amjad AI |
data-persona |
Inline system prompt & tone | You are a helpful customer support agent. |
data-persona-url |
Remote URL to a system prompt .txt / .md file |
https://site.com/prompt.md |
data-knowledge |
Inline business facts, FAQs, or pricing text | Pricing: $0/mo, Support: help@acme.com |
data-knowledge-url |
Remote URL to a live .md, .txt, or .json file |
https://site.com/knowledge.md |
data-greeting |
First message sent to visitors | Hi! How can I help you? |
data-color |
Primary brand accent color | #ef4444 or #6366f1 |
data-key |
Your ROUTER_API_KEY (if protected) |
amjadisthebest |
const response = await fetch("https://your-zeroroute.vercel.app/v1/chat/completions", {
method: "POST",
headers: {
"Content-Type": "application/json",
"Authorization": "Bearer your-router-api-key"
},
body: JSON.stringify({
stream: true,
messages: [
{ role: "system", content: "You are a helpful customer support chatbot." },
{ role: "user", content: "Hello! What are your business hours?" }
]
})
});\\python from openai import OpenAI
client = OpenAI( base_url="https://your-zeroroute.vercel.app/v1", api_key="your-router-api-key" )
response = client.chat.completions.create( model="default", # Or specify exact provider/model messages=[ {"role": "system", "content": "You are a professional assistant."}, {"role": "user", "content": "Summarize this article."} ] )
print(response.choices[0].message.content) \\
curl -N https://your-zeroroute.vercel.app/v1/chat/completions \
-H "Authorization: Bearer your-router-api-key" \
-H "Content-Type: application/json" \
-d '{"stream":true,"messages":[{"role":"user","content":"Explain quantum computing in one sentence."}]}'
\\\
---
## 🔑 How to Get Free API Keys
- **Groq Console**: [console.groq.com/keys](https://console.groq.com/keys)
- **SambaNova Cloud**: [cloud.sambanova.ai](https://cloud.sambanova.ai)
- **Mistral AI Console**: [console.mistral.ai](https://console.mistral.ai)
- **Google AI Studio**: [aistudio.google.com](https://aistudio.google.com)
- **OpenRouter**: [openrouter.ai/keys](https://openrouter.ai/keys)
- **NVIDIA NIM**: [build.nvidia.com](https://build.nvidia.com)
- **Cloudflare AI**: [dash.cloudflare.com](https://dash.cloudflare.com)
- **Cohere Dashboard**: [dashboard.cohere.com](https://dashboard.cohere.com)
---
## 🛡️ Security & Privacy
- **Encrypted Local Credentials**: Saved keys are encrypted using AES-256-GCM (secrets.json + .master.key).
- **Masked Secrets**: Keys are never transmitted in full to the client ( irst4••••••••last4).
- **Git Shield**: All credential files (.env, secrets.json, .master.key) are in .gitignore.
---
## 👨💻 Author & Creator
**Amjad P A** — Full-Stack AI Engineer & Solo Builder
- 🌐 **Portfolio**: [amjad.mapki.in](https://amjad.mapki.in)
- 🐙 **GitHub**: [@amjadlle](https://github.com/amjadlle)
- 💼 **LinkedIn**: [in/amjadlle](https://linkedin.com/in/amjadlle)
- ✉️ **Email**: [hire.amjad@gmail.com](mailto:hire.amjad@gmail.com)
---
## 📄 License
MIT License © 2026 Amjad P A. Free for personal, commercial, and open-source use.