I have a folder on my desktop called "experiments." It is full of half finished scripts, abandoned ideas, and small pieces of code I wrote on weekend mornings when the house was quiet. Most of them do nothing now. But a few of them turned into something else. They turned into small, steady deposits.

It happened by accident at first. I needed a tool that did not exist, so I built a rough version using someone else's API. Then a friend asked if he could use it. Then another person. I put up a simple page, added a payment button, and forgot about it. Years later, it still sends money every month.

I have been collecting APIs for a long time. Not the expensive enterprise ones, but the free ones, the ones with generous tiers and good documentation. Some of them have become the backbone of those small experiments that pay. Here are the ones I keep coming back to.

There is a pattern I noticed years ago. The companies that offer the best free APIs are usually the ones trying to solve a harder problem. They give away the easy stuff because they want you to need the advanced stuff later. But for someone building small things, the easy stuff is often enough.

I have built entire projects that never exceeded the free tier. The API provider probably considered me a freeloader. But I was also a proof of concept, a demonstration that their API worked, a tiny advertisement running in the background. It felt like a fair trade.

There is an API from Google that I have been using since 2019. It does speech to text. I built a small tool for myself that transcribed my voice notes and dumped them into a text file. Then a writer I knew asked if she could use it for interviews. Then her editor asked. Now there are maybe fifty people using it, and the API calls cost pennies a month because Google gives every project sixty minutes of free audio processing per month. The tool just sits there, running, depositing small payments into a Stripe account I check twice a year .

Another one is from a company that does image recognition. I used their free tier to build a script that sorts my photo library by what is in the pictures, dogs, beaches, documents. I put it on a Github repository with minimal instructions. People found it. Some of them paid me five dollars for a version that did not require the command line. The API calls are still free for the first thousand each day. Most days, the script stays well under that limit.

There is a platform called Hugging Face that hosts thousands of models you can call through an API . I do not understand most of what is on there. But I understand enough to grab a model that does one thing well, like summarizing text or generating a description from a headline.

I built a small thing last year using one of their models. It takes a long article and produces three bullet points. I use it for my own reading, send articles to an email address, get bullet points back. Then I added a public version with a simple form. A few hundred people use it now. The hosting costs are zero because the model is small and the calls are infrequent. The donations button on the site covers the domain name and a little extra.

There are newer platforms now that offer inference as a service, SiliconFlow, fal.ai, Firework AI . They give you credits when you sign up, enough to build something real and see if it works. I have been playing with fal.ai for image generation. Their free credits lasted me three months of weekend experiments. By the time I ran out, I had a small thing that made a few dollars from people who wanted custom avatars. The dollars buy more credits than I need.

I spent a month building a research tool for myself. It takes a question and searches across a dozen sources, academic papers, news sites, blog archives. The hard part was not the code. It was finding APIs that returned useful data without costing a fortune.

There is a company called Tavily that built a search API specifically for AI applications . They give you a thousand free credits a month. I hooked it up to a language model and made a simple interface. Ask a question, get a synthesized answer with sources. I shared it in a forum. A few people offered to pay for a version that saved their history. That version now runs on the free tier and the occasional five dollar subscription.

Another one is Serper, which does Google search results in clean JSON . They give twenty five hundred free queries a month. I use it for a tool that monitors what people are saying about a few topics I care about. It sends me a daily email. The emails cost nothing to send and nothing to generate.

Sometimes the data you need is not in an API. It is on a website, buried in HTML. I used to write scrapers that broke every time the site redesigned. Then I found platforms that turn websites into clean APIs.

Apify has a marketplace where people publish ready made scrapers for hundreds of sites . You can run them and get JSON back. They have a free tier with five dollars of monthly credits. I use one that scrapes job postings from a few tech boards. I built a small alert service for myself. Then I made it public. The credits cover my usage and then some.

Bright Data offers something similar but more industrial . Their free trial gave me five thousand requests to test a project that never went anywhere. But the code I wrote still runs locally, pulling data I use for personal research.

The thing about building with free APIs is that the costs stay low even when people actually use what you made. But receiving money, that used to be complicated. Payment processing, invoices, tax forms. I avoided it for years.

Then I found services that handle the money part automatically. There is a new protocol called x402 that lets you charge per API call in tiny amounts, fractions of a cent . It uses cryptocurrency behind the scenes but looks like normal code. I set it up for a small image processing tool. A user pays five cents, the tool runs, the money appears in a wallet. I transfer it to my bank when it hits fifty dollars.

Coinbase has a facilitator that makes this even simpler . You wrap your endpoint with their middleware, set a price, and they handle verification and settlement. I have a tool that converts files from one format to another. It costs one cent per conversion. Last month it made twelve dollars from people who needed a quick conversion and did not want to sign up for a monthly subscription somewhere else.

The simplest way to earn from an API is to let someone else handle the customers. Apify lets you publish your own API as what they call an Actor . You write the code, set a price, and they list it in their store with fifty thousand monthly visitors. They handle billing, hosting, scaling. You just build it and forget it.

I published a small scraper there two years ago. It collects data from a site that does not have an API. I price it at two dollars per thousand records. Some months it makes twenty dollars. Some months it makes two hundred. I have not touched the code since I published it. The money just appears.

Another marketplace is for AI models themselves. Fal.ai lets you deploy models and charge for inference . I have a friend who fine tuned a model on his own art and put it up there. People pay small amounts to generate images in his style. He sends me screenshots of the dashboard sometimes. It is not a living. But it is a small, steady thing.

I am noticing the same shape here. Build something small using free tools. Put it somewhere people can find it. Let the payment layer run in the background. The APIs do the work. The code just sits there. The deposits arrive.

I do not check the dashboards often. But when I do, I am always surprised. A few dollars from a scraper I wrote in 2021. Twelve cents from an image model someone used yesterday. A subscription payment from a person who signed up three years ago and never cancelled. It is not much. But it is something. And it happens without me.

I have a new idea I am playing with. It combines a search API that gives free credits and a summarization model that costs almost nothing. I want to build something that watches a few topics and sends me a weekly memo. Just for me. But I will probably put it up publicly, add a tip jar, see what happens.

That is how the others started. Not as businesses. As things I wanted for myself. The money came later, quietly, while I was working on the next thing.

The Free Tiers I Actually Use

One Month of Credits

The Research Stack

Marketplace Dashboard

Model Earnings (Friend's Screenshot)

The Incidental Income

61% of marketers say this is the biggest marketing shift in decades.

Get the data and trends shaping growth in 2026 with this groundbreaking state of marketing report.

Inside you’ll discover:

Download your 2026 state of marketing report today.

Get Your Report