OpenAI API pricing for WhatsApp chatbots and voice agents, for Indian businesses
As on 17 September 2026, OpenAI lists API prices in US dollars per 1 million tokens. gpt-4o-mini is $0.15 input and $0.60 output, gpt-5.6-luna is $0.20 and $1.20, and gpt-realtime audio for voice calls is $32 input and $64 output. Batch jobs cost 50% less. On Convo360 you connect your own OpenAI key and pay OpenAI directly, with zero markup.
Prices as on 17 September 2026. Figures were checked against the provider's official pages on that date. Providers change prices and rules without notice, so confirm on the official page before you budget. Convo360 adds no markup: you pay the provider directly with your own account. Convo360 is not affiliated with OpenAI. OpenAI, GPT and other product names are trademarks of their owner. Prices were read from OpenAI's official pages on 17 September 2026 in USD and may change; always confirm on the official pricing page.
- OpenAI bills per token, with separate rates for input, cached input and output, all published in USD per 1 million tokens.
- As on 17 September 2026, gpt-4o-mini is listed at $0.15 input, $0.075 cached input and $0.60 output per 1 million tokens.
- Voice calls on gpt-realtime are billed in audio tokens at $32 input and $64 output per 1 million, far above text rates.
- OpenAI's pricing page does not mention GST or a free token allowance, so neither is assumed here.
- Convo360 adds zero markup: your OpenAI key goes in Settings → Provider Keys and you pick the model per agent.
OpenAI API prices as on 17 September 2026
These figures were read from OpenAI's official API pricing page on 17 September 2026. Every price is in US dollars per 1 million tokens, on the Standard tier. The table covers only text models a business chatbot is likely to run on; the official page lists many more.
| Model | Input | Cached input | Output |
|---|---|---|---|
| gpt-5.6-terra | $2.00 | $0.20 | $12.00 |
| gpt-5.6-luna | $0.20 | $0.02 | $1.20 |
| gpt-5.4-mini | $0.75 | $0.075 | $4.50 |
| gpt-5-mini | $0.25 | $0.025 | $2.00 |
| gpt-4.1-mini | $0.40 | $0.10 | $1.60 |
| gpt-4o | $2.50 | $1.25 | $10.00 |
| gpt-4o-mini | $0.15 | $0.075 | $0.60 |
The two gpt-5.6 rows are OpenAI's short context rates. The page lists higher long context rates for that family, and a separate cache write charge ($0.25 for gpt-5.6-luna, $2.50 for gpt-5.6-terra). Convo360 is not affiliated with OpenAI. Prices can change at any time, and OpenAI's page is always the final word.
Prices for voice calls: Realtime, transcription and TTS
A phone voice agent uses audio models, and audio tokens are priced well above text tokens. As on 17 September 2026 OpenAI lists:
| Model | What it does | Published price |
|---|---|---|
| gpt-realtime | Speech in, speech out | Audio: $32.00 input, $0.40 cached, $64.00 output. Text: $4.00 input, $16.00 output |
| gpt-realtime-mini | Speech in, speech out | Audio: $10.00 input, $0.30 cached, $20.00 output. Text: $0.60 input, $2.40 output |
| gpt-realtime-2.1 | Speech in, speech out | Audio: $32.00 input, $64.00 output. Text: $4.00 input, $24.00 output |
| gpt-4o-mini-transcribe | Speech to text | Estimated $0.003 per minute |
| gpt-4o-transcribe, Whisper | Speech to text | $0.006 per minute |
| gpt-4o-mini-tts | Text to speech | $0.60 text input, $12.00 audio output |
| tts-1, tts-1-hd | Text to speech | $15.00 and $30.00 per 1 million characters |
Token prices are per 1 million tokens. OpenAI also states that the Realtime API is not priced separately: you pay the chosen model's token rates. The pricing page does not publish how many audio tokens one minute of speech uses, so we do not quote a per-minute figure for gpt-realtime. Read it from your own usage dashboard after a few test calls.
What a token is, and what you are billed for
A token is a small chunk of text. OpenAI's rule of thumb is that 1 token is about 4 characters or 0.75 words of English. Hindi, Marathi or Hinglish text can split differently, so test real replies in OpenAI's tokenizer tool.
- Input tokens: everything you send on each turn, including the system prompt, knowledge base passages, chat history and the customer's message.
- Cached input: a repeated prompt prefix billed at the lower cached rate. OpenAI describes this discount as up to 90%.
- Output tokens: the reply the model writes. These cost the most per token.
- Batch API: 50% lower cost with a 24-hour turnaround. It suits offline jobs such as tagging old chats, not live replies.
- Free allowance: the pricing page lists no free token allowance for chat models. Only omni-moderation-latest is marked Free.
OpenAI's rate limits guide places an account in Tier 1 once $5 has been paid, with a usage limit of $100 per month. Tier 2 needs $50 paid and allows $500 per month.
A per-conversation formula with illustrative numbers
Cost per conversation = (input tokens ÷ 1,000,000 × input price) + (cached tokens ÷ 1,000,000 × cached price) + (output tokens ÷ 1,000,000 × output price).
The token counts below are ILLUSTRATIVE: a WhatsApp chat of about eight messages that uses 6,000 input tokens in total and 800 output tokens, with no caching. Replace them with figures from your own usage dashboard.
| Model | Working | One chat | 1,000 chats |
|---|---|---|---|
| gpt-4o-mini | $0.0009 input + $0.00048 output | $0.00138 | $1.38 |
| gpt-5.6-luna | $0.0012 input + $0.00096 output | $0.00216 | $2.16 |
| gpt-4o | $0.015 input + $0.008 output | $0.023 | $23.00 |
For a voice call the same formula applies with audio token counts and the audio prices above. How to measure tokens properly, and how chat history inflates input, is covered in estimating your AI provider bill.
Paying OpenAI in USD from India
OpenAI's API price list is in US dollars. We do not convert it for you, because the rate that matters is the one your bank applies on the day:
₹ cost = USD amount × your bank's rate on the day + GST and forex fees, if any.
- The pricing page does not say whether taxes are included or extra. Check your first OpenAI invoice before you fix a budget.
- If you pay by card, it must be enabled for international transactions. Forex mark-up varies by bank. Ask your bank; we do not quote a figure.
- Minimum top-up, credit expiry and refund rules sit in OpenAI's billing help pages. We could not read those on 17 September 2026, so confirm them on the Billing page of your OpenAI account.
Your Convo360 subscription is a separate rupee bill with a GST invoice. OpenAI usage never passes through it.
Adding your OpenAI key in Convo360 and choosing the model
Convo360 is BYOK. You bring your own OpenAI account, OpenAI bills you directly, and Convo360 adds zero markup on tokens.
- Create an API key in your OpenAI account and add billing there.
- In Convo360, open Settings → Provider Keys, paste the key under OpenAI and click Test connection. Keys are encrypted at rest.
- Open your AI agent. Under Brain and knowledge, pick the OpenAI model in AI model (text replies). The choice is per agent, so an FAQ bot and a sales bot can run on different models.
- For a voice agent, set Speech engine (hears the caller) to OpenAI. The speaking voice can be OpenAI, Sarvam, Smallest.ai or ElevenLabs. Voice calls need the Pro plan at ₹2,999 per month, excluding 18% GST.
- Test with Preview / Try-it, then on a real WhatsApp chat or phone call.
If the key expires or the OpenAI account runs out of credit, OpenAI starts returning errors. Convo360's Provider Health monitoring detects this and shows a clear alert, so you know the fix is a top-up or a new key and not a broken flow. More detail is in the AI agents docs and in how a BYOK platform works.
Ways to cut the OpenAI bill
- Shorten the system prompt. It is re-sent on every turn, so each line you remove is saved on every message.
- Keep the prompt prefix stable. Repeated prefixes can bill at the cached rate, for example $0.075 instead of $0.15 on gpt-4o-mini.
- Send fewer knowledge base passages. Three relevant chunks cost less than ten loosely related ones.
- Match the model to the job. Because the model is set per agent, only the agent that needs a larger model pays the larger rate.
- Use guided flows where AI adds nothing. A menu, a form or an order-status lookup built with Buttons and List Menu blocks uses no tokens at all.
- Keep voice replies short. Audio output is the costliest line on the Realtime price list, so one or two sentences per turn matters.
- Check usage weekly in the OpenAI dashboard during the first month.
Weighing providers? OpenAI vs Claude vs Gemini for chatbots covers the non-price factors.
Example conversation
Connect your own OpenAI key, zero markup
Start free in minutes — no code, no credit card. Bring your own AI keys and pay no per-message markup.
Frequently asked questions
Is there a free tier on the OpenAI API?
As read on 17 September 2026, OpenAI's API pricing page lists no free token allowance for chat models; only omni-moderation-latest is marked Free. The rate limits guide shows a Free usage tier, but that is a rate limit tier, not a published grant of free tokens.
Is GST extra on OpenAI API prices in India?
OpenAI's pricing page does not mention GST or any other tax, so we do not state whether it is included. Prices are listed in US dollars. Check your first OpenAI invoice, or ask OpenAI support, before finalising a rupee budget for your chatbot.
What is the OpenAI API price per minute for a voice call?
OpenAI prices gpt-realtime per audio token, not per minute: $32 input and $64 output per 1 million audio tokens as on 17 September 2026. The page does not publish tokens per minute, so measure a few test calls in your usage dashboard.
How much does gpt-4o-mini cost per 1 million tokens?
As on 17 September 2026, OpenAI lists gpt-4o-mini at $0.15 per 1 million input tokens, $0.075 per 1 million cached input tokens and $0.60 per 1 million output tokens on the Standard tier. The Batch tier lists $0.075 input and $0.30 output.
Can I use my own OpenAI account with Convo360?
Yes. Convo360 is BYOK, so you paste your own OpenAI API key under Settings → Provider Keys, test the connection and pick the model on each agent. OpenAI bills your account directly in USD, and Convo360 adds zero markup on tokens.
What happens if my OpenAI credit runs out while the bot is live?
OpenAI starts returning errors for every request, so the AI cannot reply. Convo360's Provider Health monitoring detects the provider error and shows a clear alert, which tells you to top up the OpenAI account or replace the key rather than debug your flow.
Official sources
- OpenAI API docs: Pricing (official)
- OpenAI API docs: Prompt caching (official)
- OpenAI API docs: Batch API (official)
- OpenAI API docs: Rate limits and usage tiers (official)
- OpenAI API docs: Key concepts, tokens (official)
Related reading
Available across India — see WhatsApp & AI calling automation by state and city.