AI Path Finder
Free consultation

Lesson · Hinglish · ⏱ 8 min padhne mein

🔌 LLM API call karna: keys, requests aur costs

EnglishHinglish

🏠 Asli zindagi ki misaal

Zomato pe order karna: aap request bhejte ho (kya chahiye, kahan chahiye), restaurant use process karta hai, aur aapko response milta hai (khaana). Har order ka paisa lagta hai, aur account login batata hai ki aap kaun ho. LLM API bhi aisa hi hai: aapka code messages ke saath request bhejta hai, model jawaab deta hai, billing tokens ke hisaab se hoti hai, aur API key aapka login hai.

💡 Aasaan shabdon mein

Anthropic aur OpenAI jaise providers apne models internet pe API ke through use karne dete hain. Aap messages ki list (system aur user) aur settings jaise model ka naam, max tokens aur temperature bhejte ho, aur generated text wapas milta hai. API keys secret rakho: environment variables mein daalo, code ya GitHub mein kabhi nahi. Errors aur rate limits ke liye retries aur timeouts ka plan rakho. Streaming jawaab word by word bhejta hai, taaki user ko wait na karna pade. Cost track karo: input tokens plus output tokens, guna price per token.

🪜 Step by step

  1. Provider ke console mein API key banao aur use environment variable mein rakho, code mein kabhi nahi.
  2. Model ka naam, system message (instructions) aur user messages ke saath request bhejo; response se text padho.
  3. Input aur output token counts log karo. Cost = tokens × per token price, isse per user cost ka andaaza lagta hai.
  4. Failures sambhalo: timeout set karo, rate limit (429) aur server errors pe exponential backoff ke saath retry karo, aur sab retries fail hon to friendly message dikhao.
  5. Lambe jawaab stream karo taaki users ko turant text dikhe, aur cost control ke liye max token limit lagao.

🔑 Zaroori shabd

API key
Secret password jo API ko aapka account batata hai.
Rate limit
Ek minute mein maximum kitni requests allowed hain.
Streaming
Jawaab ko banate-banate tukdon mein receive karna.

👀 Example dekho

python
import os, anthropic
client = anthropic.Anthropic(api_key=os.environ["ANTHROPIC_API_KEY"])

msg = client.messages.create(
    model="claude-sonnet-5-5",
    max_tokens=300,
    system="You are a concise travel assistant for Indian travellers.",
    messages=[{"role": "user", "content": "3-day Jaipur plan on a budget?"}],
)
print(msg.content[0].text)
print(msg.usage)   # input/output tokens → cost

Key environment variable se aati hai, code mein kabhi paste nahi hoti.

⚠️ Aam galtiyaan jinse bachna hai

  • API keys GitHub pe commit karna. Bots minuton mein dhoondh lete hain; aisa ho to turant key badlo.
  • Browser code se API call karna, jisse key khul jaati hai. Apne server se call karo.
  • Koi usage limit nahi. Ek bug ya misuse bahut bada bill bana sakta hai; budgets aur per user limits lagao.

✅ Jaldi se check karo

Aapki API key kahan honi chahiye?

  1. Environment variable ya secret manager mein
  2. README mein
  3. app.py mein hard-coded
Jawaab dekho

A. Environment variable ya secret manager mein: Code mein rakhi keys GitHub pe pahunch kar chori ho jaati hain. Environment variables ya secrets manager use karo.

🙋 Seekhne wale yeh poochhte hain

Kis model se shuru karun?

Feature kaam karta hai yeh prove karne ke liye ek strong general model se shuru karo, phir apne eval set pe chhote, saste models test karo aur jahan quality bani rahe wahan switch karo.

Kya mera data training ke liye use hota hai?

Policies provider aur plan ke hisaab se alag hain. Business API plans aam taur pe default mein aapke data pe train nahi karte, par current terms hamesha check karo.

📌 Yaad rakho

Request andar → response bahar, billing tokens se. Key sambhaalo, aur errors handle karo.

📖 Is lesson ke shabd (AI dictionary)

Poora course, aapke roadmap ke hisaab se

Progress tracking, quick checks aur step-by-step order ke saath. Yeh lesson in courses ka hissa hai:

AI Foundations Path · AI Product Engineer · LLM / Agent Engineer · AI Forward Deployed Engineer

Course mein seekho