On July 30, 2026, OpenAI quietly cut the price of GPT-5.6 Luna by 80%. Around the same time, ChatGPT crossed roughly 1 billion weekly active users. Neither of those facts made much noise on its own, but put together they tell you something worth knowing before you decide what to actually pay for: OpenAI is fighting a price war it didn’t start, and it’s fighting it on the cheapest tier first.
Here’s exactly what got cheaper, who it actually matters for, and — separately — what the real cheapest way to use ChatGPT is if you’re just a regular user in India, because those are two different questions with two different answers.
What actually got cheaper
This price cut applies to the API — the pay-per-token pricing developers use to build apps and automations on top of GPT-5.6, not the ChatGPT app subscription most people pay for. OpenAI credited the drop to system efficiency improvements, not a change in the model itself.
| Model | Old price (in / out per 1M tokens) | New price (in / out per 1M tokens) | Change |
|---|---|---|---|
| Luna | $1.00 / $6.00 | $0.20 / $1.20 | −80% |
| Terra | $2.50 / $15.00 | $2.00 / $12.00 | −20% |
| Sol | $5.00 / $30.00 | $5.00 / $30.00 | No change |
Luna — the fastest, lightest tier in the GPT-5.6 family — is now roughly five times cheaper than it was in June. Terra got a smaller 20% cut. Sol, the flagship reasoning tier, wasn’t touched at all, which tells you where OpenAI thinks the actual price competition is happening: high-volume, everyday tasks, not the hard reasoning jobs.
Who this actually matters for
If you just use the ChatGPT app to ask questions, this change does nothing to your bill — your subscription price hasn’t moved. It matters if you’re on the other side of the API: a developer, a small business running an automated WhatsApp or Instagram bot, or anyone paying per-request through a tool built on GPT-5.6.
For high-volume, simple tasks — auto-replying to customer messages, drafting daily social captions, summarizing orders — Luna at $0.20/$1.20 per million tokens is now cheap enough that cost stops being the reason to avoid automating something. If you’re already using AI to generate daily content for a small business, this is the tier that makes running it as an always-on automation realistic instead of a manual daily chore.
So what’s the actual cheapest way to use ChatGPT?
Separate question, separate answer. If you’re a regular user in India paying for the app rather than the API, here’s where the plans actually sit right now:
| Plan | Price (India) | Best for |
|---|---|---|
| Free | ₹0 | Occasional use, no serious limits pressure |
| Go | ₹399/month | Daily use on a budget — the real “cheapest way to use ChatGPT” for most people |
| Plus | ₹1,999/month (incl. GST) | Heavier daily use, image generation, higher limits |
| Pro | ₹19,900/month | Power users running GPT-5.6 Sol at max reasoning constantly |
For most people asking “what’s the cheapest way to actually use ChatGPT,” the honest answer isn’t the Luna price cut at all — it’s the Go plan at ₹399/month, which has been available in India since last year and remains unchanged. The Luna cut is a developer story, not a consumer discount, even though it’s easy to see the “80% cheaper” headline and assume it applies to your own subscription.
Why OpenAI is cutting prices at all
This isn’t happening in a vacuum. Google’s Gemini and Anthropic’s Claude are both pushing hard on price-to-performance this year, and cheaper video and image models like the ones we covered in our Seedance vs. Veo vs. Kling vs. Runway comparison are putting pressure on margins across the entire AI industry, not just chatbots. Cutting Luna’s price first, while leaving Sol untouched, is a fairly clear signal: OpenAI is defending the high-volume, low-margin end of the market where competitors are cheapest, while protecting pricing on the tier serious reasoning work actually depends on.
Frequently asked questions
No. The 80% cut applies to API pricing — the pay-per-token cost developers pay to build apps on GPT-5.6 — not the ChatGPT app subscription (Free, Go, Plus, or Pro). Those subscription prices are unchanged.
For most everyday users, the ChatGPT Go plan at ₹399/month is the cheapest paid option with meaningful daily limits. The free plan remains ₹0 for occasional, lighter use.
Input token pricing dropped from $1.00 to $0.20 per million tokens, and output pricing dropped from $6.00 to $1.20 per million tokens — an 80% cut on both.
No. Sol, the flagship reasoning tier, kept its pricing unchanged at $5 per million input tokens and $30 per million output tokens. Only Luna (-80%) and Terra (-20%) were cut.
Mainly developers and businesses running high-volume, simple automations on the API — things like customer-message auto-replies, content drafting, or order summaries — where cost per request matters and Sol-level reasoning isn’t needed.
The bottom line
GPT-5.6 Luna getting 80% cheaper is real, but it’s a developer-side story, not a reason your own ChatGPT bill is about to drop. If you’re building something on top of GPT-5.6 — an automation, a bot, a daily content pipeline — Luna just became a lot easier to justify running constantly. If you’re just deciding which plan to pay for as a regular user, the answer hasn’t changed: Go at ₹399/month is still the cheapest sensible option, and this price cut doesn’t touch it.
Are you building anything on the API, or just using the app? Let us know in the comments, and subscribe to ournationonline for more grounded breakdowns of what these pricing changes actually mean.
