Why your Indian card keeps getting declined on OpenAI
If you've tried to pay for ChatGPT Plus, OpenAI's API, or DALL·E credits from India in the last year, you've likely seen one of these errors: 'Your card was declined,' 'Card not supported,' or the silent 'something went wrong, please try again' that costs you a half-hour of debugging.
Three things drive this. First, OpenAI's payment processor (Stripe) flags India-issued cards as elevated risk by default; some banks rate-limit these attempts. Second, Indian banks have their own international-card filters and frequently flag OpenAI as unusual merchant activity, especially on first attempt. Third, 3D Secure / OTP friction is real — your bank sends an OTP, you don't see it within 60 seconds, the auth times out, the card is declined.
The 3.5% forex you're also paying
Even when the card works, every $20 ChatGPT Plus charge picks up a 3.5% forex markup on a regular Indian card — about ₹65/month, or ₹780/year. Add Claude Pro, Cursor, GitHub Copilot, and an OpenAI API budget on top and the forex bill on AI tools alone is ₹2,500–₹4,000/year.
The fix: a virtual card outside the Indian banking system
The pattern that consistently works for ChatGPT Plus in India:
• Use a virtual Visa from a non-Indian BIN. Plu fits this — the card is issued by a US bank rail and the BIN doesn't trip Indian-card filters at OpenAI.
• Use a card with zero forex markup, so the $20 charge converts at the network rate, not the bank-markup rate.
• Use a card with no 3D Secure / OTP friction on US merchants, so the auth flow doesn't time out.
Plu hits all three. Net cost: $20/month flat, no markup, no decline retries.
Step by step: subscribe to ChatGPT Plus from India with Plu
1. Sign up at app.getplu.com — passport KYC, ~2 minutes, no Indian bank account or credit check.
2. Fund the card. UPI is fastest if you're spending INR you already have; USDC is best if you hold stablecoins. Even ₹2,000 is enough to cover one month plus headroom.
3. Issue a virtual Visa (free on the Free tier). Copy the card number, expiry, and CVV.
4. Go to chatgpt.com → settings → billing → upgrade to Plus. Paste the Plu card details. Use any Indian address — the card doesn't AVS-check the address against your bank record.
5. Confirm. The charge clears. ChatGPT Plus is active.
If you want belt-and-braces, issue a card-of-the-month: a virtual card with a $25 limit, used only for ChatGPT Plus. If anything goes wrong (subscription disputes, surprise renewals), the blast radius is bounded.
Other AI tools the same trick works for
The same setup works for any USD-billed AI service that gives Indian cards trouble:
• Claude Pro ($20/month) and Anthropic API • Cursor ($20/month) • Replit Pro / Replit Agent • Perplexity Pro • Mistral Le Chat Pro • ElevenLabs (audio) • Runway / Suno (video, music) • OpenAI API direct (you'll spend in credits, not subscription)
If you're running AI workflows or agents, issue a separate virtual card per tool with a hard monthly limit. That way Cursor can't accidentally charge you for ten months of overages, and a misbehaving agent can't drain your account.
Frequently asked questions
Why does OpenAI specifically reject Indian cards? OpenAI's processor (Stripe) and their fraud heuristics flag India-issued cards as elevated risk for chargebacks. Your bank also has filters in the other direction. The combination produces frequent declines on first attempt.
Is using a non-Indian-BIN virtual card legal in India? Yes. Plu cards comply with international card-issuance rules and operate within Visa's network. You're paying a USD merchant from a USD balance you've funded — the same as using a Wise or Revolut card abroad.
Do I need an Indian credit card to subscribe to ChatGPT Plus? No. Plu's virtual Visa works for OpenAI's checkout flow without any Indian banking involvement.
Will ChatGPT Plus charges show up in my Indian credit-card statement? Only if you fund Plu from an Indian credit card and then OpenAI charges Plu. Most users fund with UPI or USDC, in which case there's nothing on their Indian credit-card statement at all.