Hi DEV π I'm Alba, a solo iOS developer from Spain.
If you use AI tools every day, you've probably noticed how fast tokens add up, especially on paid plans and APIs.
I've built an app that, in my tests, cuts that token usage by 80β96% for the same results. It will be available for iPhone, iPad and Mac (iOS, iPadOS and macOS). It's currently in App Store review, so I'm keeping the details secret until it's approved π€«
Before I set the price, I'd love your honest opinion.
A few questions
- How much do you spend on AI tokens per month? (A rough range is perfect.)
-
If an app cut that by 80% or more, what would you pay for it?
- One-time, under $10
- One-time, $10β30
- Subscription, around $3/month
- I'd need to try it first
- What would make you trust a number like "80β96%"? A free trial, a live demo, a side-by-side comparison?
Why I'm asking
I'm an indie dev, and all my apps are privacy-first: no accounts, no tracking, everything on device. I want this one to follow the same rules and be priced fairly, not like another subscription you forget about.
I'll share everything about how it works once it's live. Follow me if you want to see the launch π
Thanks for reading, and I'll reply to every comment! π
Top comments (3)
I'd have to see it first. I spend around $150 a month on all my subscriptions and OpenRouter. If I was truly convinced that It would save me that many tokens then a three dollar subscription is an easy sell. Though, of course I would prefer a one-time fee because I am, like everyone is, really cheap right now. But if it would let you get that much more output for the same API spend, or that many more tasks done in a fixed subscription, it seems economically irresponsible to not pay $3 a month for that.
Thank you, this is exactly the kind of answer I was hoping for! π
$150/month across subscriptions and OpenRouter is a great data point. And I hear you on the one-time fee, I'm seriously considering offering both.
Quick follow-up: what would convince you first? A free trial with your own prompts, a before/after token count, or a short video demo?
I'd need to try it before paying, mainly because the 80β96% figure raises an important question: what exactly is being reduced, and how are you measuring whether the results are genuinely equivalent?
I'd trust the claim much more with a side-by-side test showing the original prompt, the optimized version, token counts, output quality, and any information lost in the process. Ideally, let users test it on their own real prompts rather than only curated examples.
On pricing, I'd lean toward a one-time purchase in the $10β30 range if the app delivers consistent savings and works locally without sending prompts through another service. A subscription would be harder to justify unless there's an ongoing infrastructure or maintenance cost.
One thing I'd be especially interested in is whether the savings apply across different models and providers, or mainly to a particular workflow.