Garmin Reopened Its Developer API. The Free Tier Has No History, and Bigger Apps Will Pay Per User From April 2027

On 21 October Garmin reopened its Connect Developer Program, the cloud route to Garmin’s Activity, Health, Women’s Health, Training and Courses data, after about seven months closed to new applicants [3]. DC Rainmaker and iDevice date the closure to March [3][10]. Garmin never announced a date, but developers’ correspondence shows applications stalling from late March and the application form gone by May [12]. The reported reason was volume: requests had gone from a few a week to a daily flood, many of them for AI-generated apps [3][10].

It came back with three tiers, a per-user fee for apps at scale and a contract that matters more than either. The short version for a product team:

• Getting in is easy again. A business can apply today and build on Starter Access for free [1][2].
• The free tier is built for building, not for onboarding at scale. Starter delivers nothing from before a user connects, has no pull route and has limited user capacity [1].
• Scale now has a price. Enterprise carries a $1,000 entry fee and a negotiated fee per active monthly user. Existing partners start paying on 1 April 2027 [3][10].
• The terms constrain your architecture more than the fee does. Garmin’s agreement limits who else may touch the data, requires attribution on anything derived from it and sets conditions on AI use [6][7].

Garmin is the second wearable maker this month to put a price on data access, after Withings made Withings+ a condition of Apple Health sync for new accounts. Our September account of the pause is here.

What did Garmin open on 21 October?

From Garmin’s own program page [1]:

Garmin’s page publishes no prices; the figures in the last row come from DC Rainmaker’s briefing [3]. The application form itself is more revealing than the page [2]. It is for “businesses, organizations and other commercial entities”, must be submitted by whoever owns or distributes the app, and requires a privacy policy URL, a business model, the expected number of users in the next 12 months (0 to 100, 100 to 1,000, or over 1,000) and a list of “any third-party services, including analytics, machine learning or AI platforms” that will process Garmin data. Its list of business types includes “Data & Integration Services” and “AI & Foundation Model Services”.

Starter apps also carry an unverified label until Garmin reviews them, and new user connections may be limited during that review [5].

What does the free Starter tier leave out?

Starter has three gaps, and the first is the one that shapes a product.

No history. With Starter, data from before the user connected is not delivered [1]. A Garmin owner who has worn a watch for three years connects to your app and starts at zero. Anything that needs a baseline has to wait for one to accumulate:

The first week is when most apps win or lose a user. A personalised product on Starter spends that week, and most of the month, without the data it needs to personalise anything. We covered how to design around that gap in cold-start personalization; on Starter it applies to Garmin users as well as new wearers.

Push only. Garmin sends data to your endpoint as the user’s device syncs. There is no ping/pull route on Starter [1], so your webhook receiver has to be public, always up and idempotent, because there is no documented way on this tier to ask Garmin for a day you missed.

Limited capacity. Garmin says “Limited” [1]; DC Rainmaker reports about 1,000 users [3]. That is enough to build, test and run a pilot. It is not enough for an employer program, an insurer, or a consumer app with any traction, which is the point: past that line, Garmin expects you on Enterprise.

How much will Garmin Enterprise cost?

As reported by DC Rainmaker and iDevice [3][10]:

• $1,000 entry fee, credited against the monthly bill.
• A monthly fee per active user, negotiated per company. No rate is published.
• Existing partners move to Enterprise automatically, skip the entry fee and start paying on 1 April 2027.

This is not Garmin’s first fee. It charged a $5,000 entry fee for API access until 2019 [3], and the developer agreement has always allowed Garmin to start charging on 30 days’ notice [6, section 10.1].

Without a published rate, the arithmetic that matters is simple: every $0.10 per active user per month is $1.20 per user per year. At 10,000 active Garmin users that is $12,000 a year; at 100,000, $120,000. Whether that is trivial or material depends on what a user is worth to you, which is why the definition of “active” is the clause to negotiate.

For a passive-data product, Garmin’s active user is probably not your active user. Garmin pushes data whenever the watch syncs, whether or not the person ever opens your app. So the bill is likely to track people who are wearing a Garmin, not people who are engaged with your product. For an employer or insurer program, where participation and engagement are different numbers, that gap is where the cost hides. Questions to settle in writing before you sign:

1. Is “active” a user who synced data that month, a user your app requested data for, or any connected user?
2. Are disconnected or dormant users excluded automatically, and when?
3. How much history does Enterprise backfill: exactly one month, or more by agreement?
4. Is the rate tiered by volume, and does it differ by API, for example Health only against Health and Activity?
5. What notice will Garmin give before changing the rate?

Is Enterprise worth paying for?

Two questions decide it for most products:

1. Will you pass about 1,000 Garmin users in the next year? Then you will need Enterprise regardless, and the only question is the price.
2. Does your product need a personal baseline in the first month? If a new user’s experience depends on 7 or 28 days of history, Enterprise’s month of backfill is the difference between a personalised first week and an empty one.

If neither applies, Starter is enough. If either does, estimate the annual cost as Garmin users who sync in a typical month × the quoted rate per active user × 12, and set it against what a Garmin user is worth to you over the same year. Garmin has not published a rate, so the formula works with whatever it quotes.

Which contract terms matter most?

Most teams read the tier table and skip the agreement. The agreement is where the constraints are. From the Garmin Connect Developer Program Agreement, Rev. B, dated March 2026 [6]:

Two practical consequences follow.

Name everyone in your pipeline. The application asks which third parties will process, analyze or store Garmin data [2], and 5.2(o) is how Garmin enforces the answer. DC Rainmaker reports that Garmin cut off one large aggregator’s access earlier this year until it followed the program’s rules [3]. If an analytics vendor, a data platform or an LLM provider touches Garmin data, list it on the form.

AI is permitted, with paperwork. Garmin is more permissive than Strava, which bans putting its data into an AI context window, and stricter than providers that say nothing. The full comparison is in can you feed wearable data to an AI.

Is Garmin removing data from activity files?

On 6 October Runalyze reported that VO2 max estimates, recovery time and performance condition no longer arrive through the Garmin connection, because “Garmin removes unknown fields” [8]. It is the second time this year. In March, Garmin told Runalyze that “undocumented messages and fields are being removed before files are sent to third-party platforms”, then rolled the change back on 19 March while it took a closer look [9].

The original file a user downloads from Garmin Connect still contains those values [8]. The removal applies to activity files delivered to third parties, not to the Health API’s daily data, which still carries VO2 max as its own field [11]. If your product parses Garmin’s FIT files for Garmin’s own training metrics, assume those fields can disappear without notice, and read the documented summaries instead.

Can an app use a Garmin Personal API key?

The Personal API lets anyone create a key for their own account in seconds [3]. Garmin describes it as access for individuals “to explore, analyze, and build personalized experiences using their own data” [1] and, as reported by Garmin Rumors, “not intended for commercial apps that serve other people” [4]. Keys last 30 days to a year, access tokens last 24 hours, and there are no refresh tokens [4].

Some users will offer to paste their key into your app. Don’t build on it: it expires, it falls outside the business terms, and it gives you none of the consent and deletion handling the business tiers expect. The more important shift is the other way round: your Garmin users can now pull their own data into their own tools, including AI assistants, without you. An app whose only value is showing Garmin data back to its owner now competes with a free key.

What to do this quarter

If you are an existing partner: you are on Enterprise and pay from 1 April 2027 [3]. Count your Garmin users who sync in a typical month, get the definition of “active” and the rate in writing, and check that every processor in your pipeline is covered by 5.2(o) and declared to Garmin.

If you are building something new: apply for Starter now; it is free, and DC Rainmaker expects approvals to be quick [3]. Design onboarding for the empty first month: phone-based signals, questions and any wearable data already on the phone fill the gap until Garmin history accumulates.

If you will pass 1,000 users: budget for Enterprise and decide whether historical backfill is worth paying for. For anything that depends on a personal baseline, it usually is.

If you rely on an aggregator for Garmin: ask whose Garmin agreement the data flows under, whether that party is approved to pass data to you, and who pays the per-user fee from April 2027. A bundled price that ignores the fee will not stay bundled.

Whatever you do: keep a second path. Many Garmin owners also sync to Apple Health or Health Connect, so on-device reads still produce a signal for them, though with less detail than the Health API.

Garmin’s reopening is good news: anyone can build on Garmin data again. But the free route now ends at about a thousand users, scale has a per-user price, and the contract decides who else may touch the data. Treat Garmin as a valuable source with a cost and an off switch, not as free infrastructure.