FreeLLMAPI Setup Guide: Run Free LLM APIs Locally

Har mahine AI API credits ke liye $20 se $50 (₹1,500 – ₹4,000+) kharch karne ki zaroorat nahi hai agar aap ek student, beginner developer ya vibe coder ho. Groq, Google AI Studio aur Cerebras jaise top AI labs free tier dete hain jisme har mahine lakho tokens bilkul muft milte hain.

Lekin dikkat ye aati hai ki alag-alag SDKs yaad rakhna, har provider ki alag key sambhalna aur daily rate limits par error aana kaam ko slow kar deta hai. FreeLLMAPI is problem ko solve karta hai — ye sabhi free tiers ko aapke laptop par ek hi OpenAI-compatible endpoint ke peeche jod deta hai.

FreeLLMAPI kya hai?

FreeLLMAPI ek open-source (MIT license) local proxy tool hai jo aapke computer par chalta hai:

• Single Unified Key: Aapko sirf ek key (freellmapi-...) milti hai. Baaki providers ki real keys kisi script ya tool mein dalne ki zaroorat nahi.
• Automatic Failover: Agar Groq ki limit khatam hoti hai, to request crash hone ke bajay proxy khud agle healthy provider (jaise Gemini ya Cerebras) par switch ho jati hai.
• OpenAI Compatible: Ye standard /v1/chat/completions format provide karta hai, jo Claude Code, Cursor, Python scripts, aur LangChain sabhi ke saath seamlessly chalta hai.
• 100% Private & Encrypted: Aapki saari API keys aapke local SQLite database mein AES-256-GCM se encrypted rehti hain. Koi bhi key bahar kisi server par nahi jati.

Zaroori Baat: Ye setup personal learning, experimentation aur vibe coding ke liye hai. Kisi bhi provider ka free tier abuse na karein aur confidential client data free models par na bhejein.

Shuru karne se pehle (Requirements)

Setup shuru karne ke liye aapke system par bas ye 3 cheezein honi chahiye:

1. Node.js 20 ya usse upar: Check karne ke liye terminal mein node -v run karein. Agar install nahi hai to nodejs.org se LTS version download karein.
2. openssl ya Node crypto: Encryption key generate karne ke liye (Mac/Linux mein pehle se hota hai, Windows par Node.js se ban jata hai).
3. Kam se kam 2-3 Free Provider Accounts:
   • Groq: Fast Llama 3 models ke liye — console.groq.com
   • Google AI Studio: Gemini 2.0 Flash ke liye — aistudio.google.com
   • Cerebras Cloud: Super-fast Llama inference ke liye — cloud.cerebras.ai

Step 1: Repo clone karo aur npm install chalao

Apna terminal kholo (Windows user: PowerShell ya Git Bash, Mac user: Terminal) aur ye teen commands ek-ek karke run karein:

git clone https://github.com/tashfeenahmed/freellmapi.git
cd freellmapi
npm install

Kya hoga: FreeLLMAPI ka project aapke folder mein download hoga aur saari required libraries install ho jayengi. Isme 1 se 2 minute ka time lag sakta hai.

Step 2: .env file banao (Encryption Key + Port)

FreeLLMAPI ko aapki provider keys encrypt karne ke liye ek 64-character hexadecimal key chahiye hoti hai. Example file copy karke key generate karein:

macOS ya Linux (Terminal) par:

cp .env.example .env
ENCRYPTION_KEY="$(openssl rand -hex 32)"
printf "ENCRYPTION_KEY=%snPORT=3001n" "$ENCRYPTION_KEY" >> .env

Windows (PowerShell) par:

Copy-Item .env.example .env
$ENCRYPTION_KEY = node -e "console.log(require(x27cryptox27).randomBytes(32).toString(x27hexx27))"
Add-Content -Path .env -Value "`nENCRYPTION_KEY=$ENCRYPTION_KEY`nPORT=3001"

Ab aapki .env file mein PORT=3001 aur ek 64-character ki random ENCRYPTION_KEY save ho chuki hai.

Dhyan Rakhein: ENCRYPTION_KEY ko kabhi delete ya modify mat karna. Agar ye key badal gayi to purani saved keys decrypt nahi ho payengi aur unhe dubara add karna padega.

Step 3: Server chalao aur dashboard kholo

Server start karne ke liye terminal mein ye command run karein:

npm run dev

Terminal ko band mat karna. Server isi terminal session mein chalta rahega.

Ab apne browser mein terminal mein dikhaya gaya URL kholo (usually http://localhost:3001 ya dev mode mein http://localhost:5173). Aapke saamne FreeLLMAPI ka clean visual dashboard open ho jayega.

Step 4: Provider keys add karo (Keys tab)

1. Dashboard ke menu mein Keys tab par click karein.
2. Upar right corner mein + Add key button dabayein.
3. Dropdown se apna provider select karein (jaise Groq, Google AI Studio, Cerebras).
4. Uski free API key paste karein aur Save karein.
5. Isi tarah baaki 2 providers ki keys bhi add kar dein.

Pro Tip: Kam se kam 3 providers add karein taaki jab ek provider ka daily quota exhaust ho, to request bina ruke agle provider par automatically forward ho sake.

Step 5: Unified API key copy karo

Dashboard ke Keys tab mein upar aapko ek Unified API Key dikhegi jo freellmapi-... se shuru hoti hai.

Is key ko copy karke apne paas safe rakh lein. Ab se aapke kisi bhi tool, script ya coding agent mein sirf yahi ek single key use hogi.

Step 6: Claude Code (ya apne app) mein connect karo

FreeLLMAPI Anthropic Messages API (/v1/messages) ko bhi support karta hai, isliye aap terminal coding agents jaise Claude Code ko bina credit card ke chala sakte hain!

Tarika A: Environment Variables (Sabse aasaan)

Mac / Linux Terminal:

export ANTHROPIC_BASE_URL=http://localhost:3001
export ANTHROPIC_AUTH_TOKEN=freellmapi-YOUR_UNIFIED_KEY
claude

Windows PowerShell:

$env:ANTHROPIC_BASE_URL="http://localhost:3001"
$env:ANTHROPIC_AUTH_TOKEN="freellmapi-YOUR_UNIFIED_KEY"
claude

Tarika B: Automated Helper Command

npx freellmapi setup-claude --url http://localhost:3001 --api-key freellmapi-YOUR_UNIFIED_KEY

Tarika C: Apne Python Code mein (OpenAI SDK)

from openai import OpenAI

client = OpenAI(
    base_url="http://localhost:3001/v1",
    api_key="freellmapi-YOUR_UNIFIED_KEY"
)

response = client.chat.completions.create(
    model="auto",  # Proxy automatically healthy free model select karega
    messages=[
        {"role": "user", "content": "Explain binary search in 2 lines."}
    ]
)

print(response.choices[0].message.content)

Step 7: Routing “Balanced” rakho aur Test karo

Dashboard mein routing policy ko Balanced par set rakhein. Isse proxy latency aur daily token quota ke beech perfect balance banakar request send karti hai.

Setup verify karne ke liye dusre terminal window mein ye command chala kar dekhein:

curl http://localhost:3001/v1/chat/completions 
  -H "Authorization: Bearer freellmapi-YOUR_UNIFIED_KEY" 
  -H "Content-Type: application/json" 
  -d x27{"model":"auto","messages":[{"role":"user","content":"Namaste!"}]}x27

Agar JSON format mein AI ka answer mil gaya, to aapka FreeLLMAPI setup 100% successful ho chuka hai!

Problem aaye to (Troubleshooting Table)

Seedhi baat (Limits & Security)

• Models capability: Free models (Llama 3.3 70B, Gemini 2.0 Flash) testing aur general coding ke liye best hain, lekin GPT-4o ya Claude 3.5 Sonnet jaise heavy commercial reasoning ke barabar nahi hain.
• Machine dependence: Ye proxy aapke laptop par chalti hai. Laptop sleep ya band hoga to proxy bhi pause ho jayegi.
• Privacy: Apne personal passwords, sensitive client codebase ya private API secrets free tier models par kabhi na bhejein.

Guide by @sikhodigitally_ (Instagram). Daily free AI tools aur vibe coding tips ke liye follow karein!