How long do indie games stay in Early Access?

Of 1,206 paid indie games that opened in Steam Early Access from January 2020 to December 2022 with at least 10 reviews in their first year, half had released 1.0 within 24 months. The ones that left took a median 12 months.

And 423 of them, 35%, were still in Early Access when we checked on October 16, 2026, between nearly four and nearly seven years after they opened.

This is for the developer with a game in Early Access who is trying to decide when to call it done, or who has been in longer than planned and wants to know what normal looks like. Every number comes from Steam’s public store data, and the method is at the end.

Valve sets no time limit, but Steam starts counting at 12 months

Nothing in Valve’s rules ends Early Access. A game can stay in it for as long as it’s on sale. Two things put pressure on the clock instead.

First, Steam adds a notice to an Early Access store page once the game goes more than 12 months without a new build on its default branch, or without an update post. On September 15, 2026, 61% of released Early Access games showed it.

Second, since February 2026 Steamworks lets you set a planned 1.0 date, either a specific day or a looser window. It shows on your store page and in players’ Steam calendars, and it’s optional. An older answer on Valve’s Early Access page still says the field doesn’t exist; the newer section describes it.

Valve’s store questionnaire also asks every Early Access game, “Approximately how long will this game be in Early Access?” Players read that answer. The figures below are what a realistic one looks like.

Half were out within two years, and a third never left

A fifth of these games released 1.0 within six months of opening. A third had by month 12, and half by month 24. After that the line flattens.

1. Within 6 months20.0%
2. Within 12 months33.6%
3. Within 18 months43.2%
4. Within 24 months50.1%
5. Within 36 months56.3%
6. By October 2026 (45+ months)64.9%

1,206 paid games under Steam's Indie tag that opened in Early Access January 2020 to December 2022, with 10+ Steam-purchase reviews in their first 12 months. Early Access status and dates read from Steam's store service on October 16, 2026.

The 423 still in Early Access aren’t all abandoned. 273 of them drew at least one Steam review between September 2025 and August 2026, and 86 drew 12 or more. People are still finding and buying them. They’re just buying a game that says it isn’t finished.

Launch year made almost no difference. 65.8% of the 2020 openers were out by October 2026, 64.7% of 2021’s and 63.9% of 2022’s. A game that hadn’t left by its third year mostly wasn’t going to.

Still in Early Access at two years, 3 in 10 later reached 1.0

The odds of a 1.0 fall with every year a game stays in.

1. Still in at 12 months (801 games)47.2%
2. Still in at 24 months (602 games)29.7%
3. Still in at 36 months (527 games)19.7%

Same 1,206 games. Each bar counts the games still in Early Access at that age, and how many of them had released 1.0 by October 16, 2026. Every game had at least 45 months on Steam by then.

If your game is past its second anniversary in Early Access, you’re in the group that leaves less often than it stays. That isn’t a verdict on the game. It is a reason to put a real date on 1.0, or to decide on one of the other exits below, rather than let the decision make itself.

Bigger Early Access games stayed longer

The games with more players stayed in longer, and were more likely to finish.

Reviews in the first 12 months	Games	Out of Early Access by Oct 2026	Median months to 1.0
10 to 29	454	59.0%	9
30 to 99	301	59.8%	10
100 to 299	186	72.6%	14
300 or more	265	75.5%	15

That gap explains why the published figures disagree.

How To Market A Game, using VG Insights data on Early Access games since 2014 with at least 10 reviews, found a median of 260 days for games that left. GameDiscoverCo, looking only at games that sold more than 5,000 copies in their first 30 days, found a median of 437 days for 2025’s graduates and 15 months for 2026’s so far.

GameDiscoverCo’s sample is the successful end, and its medians match our bigger games, 14 and 15 months. How To Market A Game’s 260 days covers every year since 2014 and sits closer to our smaller games.

All three medians count only games that left. None of them can see the third that hadn’t.

The 1.0 month ran a median 3.4 times the months before it

For the 693 games we could measure, reviews in the 1.0 month and the month after ran a median 3.4 times the pace of the three months before. 69% at least doubled. That is Steam’s launch visibility doing its job: a game gets its full release visibility only at 1.0, and Steam emails everyone who has it wishlisted when it leaves Early Access.

It still rarely beat the Early Access launch. Only 29% drew more reviews in the 1.0 month and the next than in their first two months of Early Access, and the median 1.0 window was about half the size of the Early Access launch window.

That matches what GameDiscoverCo found from revenue estimates: 20% of 2025’s graduates and 21% of 2026’s earned more in their first 30 days at 1.0 than in their first 30 days of Early Access. Two different measures, the same conclusion. Most of an Early Access game’s launch happens once, at the start.

The year after 1.0 was usually better than the year before it. Of 357 games that spent at least a year in Early Access and released 1.0 by August 2025, 57% drew more reviews in the 12 months after 1.0 than in the 12 before, a median 1.23 times as many.

Games that took longer had a bigger lift, from a lower base.

Time in Early Access	Games measured	Median 1.0 lift	Beat the Early Access launch	Year after 1.0 vs year before
Up to 12 months	324	2.3x	24.7%	too few to measure
13 to 24 months	199	4.5x	34.7%	1.17x
25 months or more	170	5.2x	29.4%	1.76x

The lift is bigger for long stays mostly because the months before 1.0 had gone quiet, so the base is smaller. And the games that spent two years or more in Early Access and still shipped are the ones whose developers kept going. This shows that a late 1.0 still works. It doesn’t show that waiting makes it work better.

When shipping 1.0 now is the wrong move

1.0 is the one launch an Early Access game has left. Valve says it can’t turn Early Access back on for a game that has left, so spend it deliberately.

• If the core of the game is still changing, it isn’t 1.0. Players read 1.0 as finished, and the reviews from that launch will greet everyone who comes later.
• If the wishlist has stopped growing, a few months of updates may help more than a date. Steam’s update visibility rounds are shared between Early Access and after release, so Major Updates in Early Access spend some of them.
• If your price is going up at 1.0, plan the order. Valve’s launch discount doesn’t apply if the price changed in the 30 days before 1.0. Valve’s suggested options include raising the price more than 30 days ahead and discounting at launch, or raising it at 1.0 with no discount.
• If your review score is weak, fix the game first. More visibility shows more people something they didn’t like. Our review threshold study found that later reviews usually score lower than year one, not higher.

What to do if your game is in Early Access now

1. Answer the questionnaire honestly. Among games that left, the median was 12 months and a quarter took more than 23. If your answer has been out of date for a while, update it, and post a News event saying why.
2. Set the planned 1.0 date only when you believe it. A quarter or a year is still useful to players. A specific day you then miss is worse than none.
3. Keep both of Steam’s clocks running. A build on the default branch and an update-type post every year keep the not-updated notice off your page.
4. Treat 1.0 as a launch. It gets launch visibility and the wishlist email. Plan its trailer, its price and its press like a release, because to most of Steam it is one.
5. Give it a deadline. At two years, 3 in 10 games still in Early Access ever left. Decide now what you’ll do if you get there.
6. If you can’t finish it, you still have choices. Valve offers two exits: release the game as it stands, if it’s playable and well received, or retire it from the store, after which Valve says it won’t republish it and may offer refunds. A game’s reviews, wishlists and owners also stay with it if it changes hands, and some partners take on released and Early Access games. Strike Two is one of them, so weigh that knowing where it comes from.

How we measured it

The games come from Steam’s US store search under the Indie tag (listing pulled September 16, 2026). For each, Steam’s store service (IStoreBrowseService/GetItems) gave its Early Access status, its release date and, for games that have left, its original store date, read on October 16, 2026. A game counts as having opened in Early Access if it is in Early Access now, or if its original store date comes before its release date and it had Steam reviews before the release month.

28 games with no original date but more than five reviews before their release month were dated from their first review, the rule we used in our quiet launch study.

The cohort is the games that opened in Early Access between January 2020 and December 2022, had a US price above zero on September 21, 2026, and drew at least 10 Steam-purchase reviews in their first 12 months. Of 2,467 Early Access openers in those years, 353 were free and 908 had fewer than 10 reviews, leaving 1,206.

Time in Early Access is counted in calendar months, from the opening month to the 1.0 month.

Review counts come from Steam’s review histogram, which counts reviews from Steam purchases by the month each was written, read September 16 and 21, 2026, through August 2026.

The 1.0 lift compares the 1.0 month and the next with the three months before, scaled to two months, for games that spent at least three months in Early Access and released 1.0 by July 2026.

We checked the dating against Steam’s own flag on each review, written_during_early_access, on October 16, 2026.

In 25 randomly chosen games that left, the first flagged review fell in the month we dated the opening, or the month after, in all 25, and none had a flagged review after the month we dated its 1.0.

In 25 still in Early Access, every one of the 20 newest reviews was flagged. In 40 games we treated as launching straight to 1.0, none of their newest 300 reviews was flagged.

The limits: reviews stand in for sales and aren’t sales. Games removed from the store, games that went free and games not tagged Indie are missing, so the true share that never left may be higher.

Calendar months make short stays look up to a month longer or shorter than they were. And the comparison by time in Early Access describes the games that shipped, not what would have happened if they had shipped sooner.