Free & Open-Source Fitness APIs (2026)

A genuinely free fitness API usually means one of two very different things, and the difference decides your whole architecture. Either you get open data you own and self-host (wger, the free-exercise-db dataset, USDA FoodData Central, Open Food Facts) or you get the free tier of a paid product you borrow until you scale. Both cost zero dollars today. Only one of them still costs zero when you have real traffic. This page maps the genuinely-free/open options across exercise content, nutrition, and platform data, then gets honest about where "free" actually bites.

How to judge a "free" fitness API

Price is the easy part — it is zero. The questions that actually matter are the ones that show up after you ship:

• Do you own the data, or borrow it? Open datasets and self-hostable projects (wger, free-exercise-db) let you download, host, and keep the data. Free tiers of commercial APIs are access you can lose the moment you exceed a quota or the vendor changes terms.
• What are the licensing obligations? "Free" is not the same as "no strings". Public-domain sources (free-exercise-db, USDA) are effectively obligation-free. Copyleft and share-alike licenses (AGPL-3.0, CC-BY-SA, ODbL) are free to use but require attribution and, in some cases, that you open-source what you build on top or release derived databases as open data.
• Is it platform-bound? Apple HealthKit and Google Health Connect cost nothing, but they only run on one operating system and only see on-device data. Free, but boxed in.
• What breaks at scale? Rate limits, no SLA or support, feature/media gating, and data freshness. A free tier is built for prototyping, not production load.

The genuinely free/open options at a glance

All counts, rate limits, and tier details above are volatile — treat them as "as of 2026, verify in the docs."

The genuinely free and open options

free-exercise-db (yuhonas/free-exercise-db)

The lowest-friction way to bootstrap an exercise library. It is a public-domain dataset — 800+ exercises as structured JSON with fields like force, level, mechanic, equipment, primaryMuscles, instructions, and static image paths — released under the Unlicense, so there is effectively no attribution requirement or restriction at all. You clone the repo or fetch the combined exercises.json and host it yourself. The trade-off: it is a static dataset, not a maintained live service, so you own updates, hosting, and search, and the media is static images only (no GIFs or video). Freshness depends on community commits.

wger (wger-project/wger)

A self-hosted FLOSS fitness, nutrition, and weight tracker that also exposes an open exercise database over a documented REST API. Reported at roughly 845+ exercises (count varies by version and language — verify) with multilingual names, muscle groups, equipment, images, and per-entry license metadata. Public exercise-list endpoints are readable without authentication, and you can either query the hosted wger.de instance or run your own via Docker. Best for teams that want full control and an auditable open-source stack with no per-request fees. The trade-off is licensing: the software is AGPL-3.0, so copyleft applies to modifications you deploy, and the data is Creative Commons (attribution/share-alike per entry). The catalog is also smaller and lighter on media than commercial GIF libraries.

Free nutrition data: USDA FoodData Central and Open Food Facts

If your fitness product also needs food data, two sources are genuinely free rather than free-tier. USDA FoodData Central is U.S. government nutrient data in the public domain (CC0) — no license fee, just a free data.gov key — and is the authoritative choice for lab-grade nutrient values, though it is a reference database, not a food-logging engine. Open Food Facts is crowdsourced open data for packaged products and barcodes, with no API key required (only a custom User-Agent). Their catch is licensing shape, not price: USDA is obligation-free public domain, while Open Food Facts is copyleft (ODbL/DbCL, images CC-BY-SA) with attribution and share-alike strings. For the full breakdown against the commercial options, see our best nutrition and food database APIs guide.

Apple HealthKit and Google Health Connect

Free, but platform-bound — this is the trade-off in one line. HealthKit is Apple's on-device health and fitness store (iOS, iPadOS, watchOS, visionOS) and Health Connect is Google's on-device equivalent for Android. Neither charges a usage fee; the cost is that each runs inside one operating system and exposes only what is on that device, using per-data-type OS permission prompts rather than OAuth. That gives you rich, high-frequency data with tight OS-level privacy, but no server-to-server or cross-user access on its own. Two specifics to plan around: with HealthKit an app cannot tell whether read permission was denied (denied data simply appears empty), and Health Connect caps historical reads at roughly 30 days by default and adds a Play Store health-permission review step before launch. Note also that the older Google Fit REST API is being shut down (supported only through the end of 2026, with no signups since May 2024) — do not start new work on it; Google points on-device use to Health Connect.

Free tiers of commercial APIs — borrowed, not owned

The other kind of "free" is a free tier of a paid product. These are worth knowing, but understand what you are getting: access, not ownership. On the exercise side, the commercial ExerciseDB on RapidAPI is freemium — a free/BASIC tier exists but is rate-capped, and the richer catalog and animated GIF/video media (the whole selling point) effectively require paying. API Ninjas has a free tier for some endpoints, but its Exercises endpoint is reported as premium-only (Business/Professional plans), so there is effectively no free access to exercises there despite the brand having a free tier elsewhere.

On the nutrition side, Nutritionix, Edamam, and Spoonacular are all commercial/freemium. Edamam offers a free developer tier but requires attribution (link plus logo) on lower plans and keys each of its three APIs separately. Spoonacular runs a daily "points" quota where result-heavy calls cost more. Reports even conflict on whether Nutritionix still has a usable free tier — confirm directly. For cross-device wearable data, the aggregators (Terra, Vital/Junction, Rook, Spike) are subscription or usage-based from the start; they are not a free path, though some offer sandboxes for building.

The real catch of "free"

Whichever kind of free you pick, budget for these five realities:

1. Rate limits. Free and BASIC tiers enforce hard hourly and monthly request caps — fine for a prototype, not for production traffic. Even the open, hosted APIs (USDA) meter per hour.
2. No support or SLA. Free tiers carry no uptime guarantee. If it goes down at 2am, that is your problem to route around.
3. Feature and media gating. On commercial free tiers, the higher exercise counts and the GIF/video media — the reason you wanted the API — usually sit behind the paywall.
4. Attribution and copyleft obligations. "Free open data" still carries strings except for public-domain sources. wger's CC data needs attribution and share-alike; AGPL projects (wger software, the exercisedb.dev project) impose copyleft on derivatives you deploy; Open Food Facts' ODbL can require you to open derived databases. Only the Unlicense (free-exercise-db) and CC0 (USDA) are effectively obligation-free.
5. Data freshness and lock-in. Hosted APIs can change counts, schema, or pricing unilaterally, and a free tier is the first thing a vendor tightens. Open datasets flip the risk: no lock-in, but they are only as fresh as their last community commit — you own maintenance.

Which should you pick?

• If you want zero cost and zero licensing friction to bootstrap → free-exercise-db. Public domain, no auth, no attribution. You accept static images and self-maintenance.
• If you want a maintained, self-hosted stack you control → wger. Open source, self-hostable, public read API — just plan for AGPL copyleft on your deployed changes and CC attribution on the data.
• If you need authoritative nutrient data → USDA FoodData Central. CC0 public domain, license-free, lab-grade. You build the logging UX around it.
• If you need global packaged-product or barcode data → Open Food Facts. Free, keyless, huge coverage — mind crowdsourced accuracy and the ODbL share-alike obligation.
• If you need rich on-device wearable data and are single-platform → HealthKit (iOS) or Health Connect (Android). Free, deep, but boxed to one OS and on-device only.
• If you need turnkey GIF/video-rich exercise content or cross-device wearable aggregation → budget for a paid tier. The free options either lack the media or are platform-bound; commercial ExerciseDB tiers and the aggregators solve those problems for money, not for free.

Watch the fine print

Every count, rate limit, price, and tier gate above is volatile — vendors and projects change them often. Before you commit, verify the current numbers and, just as important, the current license terms at each project's primary docs. The licensing obligations (AGPL, CC-BY-SA, ODbL, CC0, Unlicense) are the part that outlives any pricing change, so read those carefully against what you intend to build and ship.