How do I tell if a free tier is honest?

Check three things: (1) Does signup require a credit card? If yes, it's a trial not a free tier — you'll be charged when you forget to cancel. (2) Is the rate limit viable for your use case, or is it designed to force an upgrade? 50 calls/day sounds free but isn't usable for anything. (3) Who funds the free tier? Open-source/government APIs have durable free tiers; VC-backed commercial APIs treat the free tier as a customer acquisition cost that gets cut when the business needs to show revenue.

Are there APIs that are truly free forever?

Open-Meteo, Frankfurter (ECB exchange rates), the Hacker News API, NOAA Weather, Nominatim, and Wikipedia REST are all backed by public institutions or open-source communities. None of them have a profit motive that would drive them to remove the free tier. Frankfurter is literally publishing ECB data — that data will be free as long as the ECB exists. These are as close to 'free forever' as exists on the internet.

Why don't all APIs offer free tiers?

Real-time data is expensive to serve at scale — financial data vendors pay exchange fees, weather APIs run expensive models, news APIs pay licensing fees for source content. The APIs that can offer genuinely free tiers are ones where the underlying data is either free (open government data, open-source) or where the marginal cost per query is essentially zero. A startup offering AI-powered data processing can't give it away indefinitely.

The Truth About Free API Tiers: Why 'Free' Often Isn't

April 22, 2026 · analysis

A taxonomy of free tiers: truly free (no key), free-with-friction (key required), and the bait-and-switch pattern.

Not all free API tiers are equal. After cataloguing 77 free APIs and scoring each on its free-tier quality, three distinct tiers emerge: truly keyless (our score: 1.0), free-with-friction (score: 0.7), and bait-and-switch (score: 0.3 or dead). The tier you're actually in determines the real cost of using an API.

This is an opinion piece based on building and monitoring this directory — your mileage will vary, but I'm fairly confident the taxonomy holds.

Tier 1: Truly Free (Score 1.0) — No Key, No Card, No Friction

An API earns a 1.0 score on our free-tier metric if it requires neither a key nor a credit card and has a usable rate limit. Examples from our seed: Open-Meteo (10,000 calls/day), Frankfurter (unlimited), Hacker News API (unlimited), Coinpaprika (25,000 calls/month), Mempool.space (unlimited), NOAA Weather (unlimited, US only).

These are genuinely free in every meaningful sense. You can integrate them in a hackathon at midnight with no account setup. They work in school environments where you can't hand out credit card details. They work in automated scripts without secret management.

What makes these sustainable? They're either government-funded (NOAA, ECB/Frankfurter, MET Norway), powered by open-source data and community infrastructure (OpenStreetMap/Nominatim, Hacker News/Firebase), or operated by projects where the API is the product itself rather than a loss-leader (Mempool.space, Open-Meteo which accepts donations).

Tier 2: Free-With-Friction (Score 0.7) — Key Required, No Credit Card

The majority of our 77 APIs fall here. You need an account and an API key, but no credit card. Examples: OpenWeatherMap (1,000 calls/day), CoinGecko Demo plan (10–30 calls/min), The Guardian Open Platform (500 calls/day), NewsData.io (200 calls/day), Alpha Vantage (25 calls/day).

The friction here is real but manageable. Signing up for a key takes 2–5 minutes. The key itself becomes a deployment complexity — you need to store it as a secret in your CI/CD pipeline, your environment variables, your app config. For teams and open-source projects, this adds meaningful overhead.

The rate limits vary widely. OpenWeatherMap's 1,000 calls/day is genuinely usable for a small production app. Alpha Vantage's 25 calls/day is not — it's designed to demonstrate the API, not run a real application. When evaluating a tier-0.7 API, check whether the rate limit is actually viable for your use case or whether it's calculated to force an upgrade.

Tier 3: The Bait-and-Switch Pattern

This is where it gets dishonest. The bait-and-switch typically works like this: launch with a generous free tier, build developer adoption, then progressively restrict or eliminate the free tier as the company needs revenue.

The clearest examples in our graveyard: Twitter (free → $100/month with almost no notice), Dark Sky (free → killed entirely when acquired by Apple), Yahoo Finance (never officially free, but the de-facto free tier was silently withdrawn). These aren't unique — they're the typical lifecycle of a VC-backed API.

A subtler bait-and-switch is the credit-card-required 'free trial'. This isn't a free tier — it's a trial that converts to paid. Weatherstack's 'free' plan actually requires a credit card at sign-up in some regions. If you see 'free trial' rather than 'free plan', assume you'll be charged.

Our Methodology

The free_tier_score in our seed data (1.0, 0.7, 0.3) maps to these three tiers. It's one input into our composite score alongside uptime and response time. The full methodology is at /methodology.

The 'no-key' collection at /no-key shows all tier-1.0 APIs we track. The 'no credit card' collection at /no-credit-card shows all tier-1.0 and tier-0.7 APIs. If you're starting a new project and want to minimise future risk, start with tier-1.0 APIs and treat everything else as potentially temporary.