2,739 games launched on Steam in September 2026, and in the median launch with at least 10 early reviews, 6.5% of first-14-day reviews mention a bug, crash, stutter or freeze. A quarter of launches sit above 12.7%. 22.1% of them announced a patch within 24 hours and 73.3% within a week. This is edition 1 of the Steam Launch Bug Index, built from public Steam endpoints.
What does the first edition cover?
The index looks at every game with a September 2026 release date on Steam, then at the subset that got reviews. Of 2,739 releases, 923 had a review summary on the store search page (Steam shows one from 10 reviews). Of those, 592 had at least 10 English reviews inside their first 14 days and make up the statistics below; the other 331 are in the CSV with their metadata only. All numbers come from the Steam launches in numbers page, which holds the full tables, the method and the downloads, and the QA hours calculator turns the genre medians into testing hours for your own build.
| Measure | September 2026 releases | n |
|---|---|---|
| Games released on Steam | 2,739 | all |
| With a review summary (10+ reviews) | 923 | 923 |
| With 10+ English reviews in the first 14 days | 592 | 592 |
| Median share of early reviews mentioning a bug | 6.5% (Q1 1.6%, Q3 12.7%) | 592 |
| Launches with 30% or more bug mentions | 4.6% | 592 |
| Median share of negative early reviews | 11.8% | 592 |
| Patch announced within 24 hours | 22.1% | 592 |
| Patch announced within 7 days | 73.3% | 592 |
| Patch announced within 30 days | 84.6% | 592 |
| Median time to first patch (patched games) | 42.8 hours | 501 |
| Games with an AI Generated Content Disclosure | 24.8% | 147 of 592 |
Source for every row: Steam Launch Bug Index, October 2026 edition, computed on 27 October 2026 from the Steam store search, appdetails, store pages, IUserReviewsService/GetAppReviews and ISteamNews/GetNewsForApp.
Which genres had the most bug mentions?
A game counts in every genre tag it carries, and genres with fewer than 8 games are left out. Simulation launches lead with a median of 8.2%, and they also patch fastest: 31% announced a fix within 24 hours. Casual games sit at the bottom with 5.3%.
| Genre | n | Median bug mentions | Q1 to Q3 | Patch within 24 h | Patch within 7 days |
|---|---|---|---|---|---|
| Simulation | 239 | 8.2% | 3.7% to 15.0% | 31.0% | 84.9% |
| Racing | 15 | 7.7% | 3.4% to 14.0% | 20.0% | 66.7% |
| Action | 206 | 7.3% | 2.3% to 13.7% | 20.4% | 69.9% |
| Sports | 22 | 7.2% | 3.2% to 13.0% | 31.8% | 77.3% |
| RPG | 134 | 7.1% | 2.3% to 14.4% | 20.9% | 80.6% |
| Free To Play | 55 | 6.7% | 0.0% to 11.8% | 9.1% | 56.4% |
| Adventure | 214 | 6.5% | 1.8% to 13.6% | 23.4% | 72.4% |
| Strategy | 148 | 6.3% | 1.8% to 13.0% | 24.3% | 80.4% |
| Indie | 417 | 6.2% | 0.8% to 12.2% | 22.5% | 74.6% |
| Casual | 245 | 5.3% | 0.0% to 11.1% | 23.7% | 70.2% |
| Massively Multiplayer | 19 | 5.0% | 1.9% to 12.5% | 21.1% | 63.2% |
The spread inside a genre is wider than the gap between genres. A Simulation game at the first quartile (3.7%) has fewer bug mentions than a Casual game at the third (11.1%). Genre tells you where to look first, not how your build will do. If your demo or hotfix needs a fresh pair of eyes before it meets that spread, a one-day Smoke Test puts one tester on it for 8 hours and returns a prioritized report in 2 business days.
Do Early Access launches get more bug mentions?
Yes, and by a wide margin. The 83 Early Access launches in the sample have a median bug mention rate of 11.1%, against 5.9% for the 509 full releases. Their negative share is similar (12.7% vs 11.5%), which fits how Early Access players behave: they report problems and still recommend the game.
| Group | n | Median bug mentions | Q1 to Q3 | Median negative | Patch within 24 h | Patch within 7 days |
|---|---|---|---|---|---|---|
| Early Access | 83 | 11.1% | 5.0% to 18.8% | 12.7% | 32.5% | 83.1% |
| Full release | 509 | 5.9% | 0.9% to 11.8% | 11.5% | 20.4% | 71.7% |
Early Access teams also patch faster: a third within a day, five in six within a week. What Valve expects from an Early Access launch, and the checklist for it, is in Early Access launch checklist: 35 checks before Steam.
Does size change the picture?
Size here is total reviews at the time of the scrape, a proxy for audience and studio size. The medians are close: 6.5% for small launches (10 to 99 reviews), 6.7% for mid (100 to 999) and 5.0% for large (1,000 and more). Large launches have the highest negative share (18.7%) and the slowest first patch (median 62.2 hours), while mid-sized launches patch fastest (37.3 hours, and 32.3% within 24 hours).
| Size | Reviews | n | Median bug mentions | Median negative | Patch within 24 h | Median time to first patch |
|---|---|---|---|---|---|---|
| Small | 10 to 99 | 436 | 6.5% | 11.1% | 19.3% | 43.2 h |
| Mid | 100 to 999 | 127 | 6.7% | 12.3% | 32.3% | 37.3 h |
| Large | 1,000+ | 29 | 5.0% | 18.7% | 20.7% | 62.2 h |
With 29 large launches in one month, the large row moves a lot between editions. Treat it as a first reading.
AI disclosure vs the rest: what does the data say?
Since 10 January 2024 Steam has asked developers to disclose AI-generated content, and the store page shows an “AI Generated Content Disclosure” section when they do (Steam news post “AI Content on Steam”, accessed 27 October 2026). In September 2026, 147 of the 592 launches in the sample carried one, 24.8%.
| Group | n | Median bug mentions | Q1 to Q3 | Median negative | Patch within 24 h | Patch within 7 days |
|---|---|---|---|---|---|---|
| AI disclosure present | 147 | 8.0% | 0.0% to 12.9% | 19.4% | 21.1% | 82.3% |
| No disclosure | 445 | 6.1% | 1.7% to 12.6% | 9.4% | 22.5% | 70.3% |
Three readings, all careful. First, the bug mention medians differ by 1.9 points, inside the month-to-month noise we expect; the quartiles overlap almost completely. Second, the negative share is twice as high in the disclosure group (19.4% vs 9.4%), and reviews mention many things besides bugs, so we cannot say from this data whether the gap is about quality or about how players react to the disclosure itself. Third, disclosed games patch more within a week (82.3% vs 70.3%). The disclosure texts themselves range from “AI-assisted translation pass reviewed by a human” to placeholder art and generated voices; the glossary entry explains what the field is. We measure this split every month and do not draw a verdict from one edition.
How fast do launches patch?
84.6% of launches announced at least one patch, fix or update within 30 days, and the median patched game posted its first one 42.8 hours after the release date. The median game posted 3 patch announcements in its first 30 days. Patch day is a normal part of a Steam launch, which is the argument for a regression pass before every one of them; the timing by genre and size, and what the first patches tend to fix, is in Day one patch: 22% of Steam launches patch within 24 hours. Whether more patches go with worse reviews, once bug mentions are taken into account, is in patch count vs Steam reviews.
Which September launches had the most and fewest bug mentions?
Only games with at least 200 total reviews are named. A high rate means players talked about bugs in the first two weeks, which often goes together with fast patching, as the first-patch column shows.
| Most bug mentions | Bug mentions | Negative | Reviews in window | AI disclosure | First patch |
|---|---|---|---|---|---|
| Police Chief Simulator | 42.0% | 32.0% | 300 | no | 17.0 h |
| Wild West Pioneers | 35.6% | 24.6% | 236 | no | 11.8 h |
| Tiny Eden | 33.3% | 26.7% | 300 | no | 22.8 h |
| Train Sim World 7 | 29.6% | 34.0% | 247 | yes | 14.0 h |
| Crimson Moon | 27.7% | 41.3% | 300 | yes | 22.2 h |
| Fewest bug mentions | Bug mentions | Reviews in window | Total reviews |
|---|---|---|---|
| 20 More, Somehow Even Smaller, Mazes | 0.3% | 300 | 1,575 |
| Normal Golf Game | 0.3% | 300 | 1,189 |
| Worming from Home | 0.3% | 300 | 798 |
| lily’s world XD | 1.0% | 300 | 1,025 |
| BOMBANANA! | 1.0% | 300 | 851 |
The full top 10 of both lists is on the Steam launches in numbers page, with Steam store links.
How was this measured, and what are the limits?
- Release list. Steam store search sorted by release date, games only, for every day of September 2026, read on 27 October 2026.
- Metadata.
store.steampowered.com/api/appdetails: type, genres, release date, review totals. - AI disclosure. The store page HTML, section “AI Generated Content Disclosure”.
- Reviews.
IUserReviewsService/GetAppReviews, English, recent first, limited to the 14 days after the release date, up to 300 per game. A review counts as a bug mention if it matches a fixed word list (bug, crash, stutter, desync, freeze, glitch, softlock, broken and their forms). - Patches.
ISteamNews/GetNewsForApp, community announcements: the first post after release whose title or first 400 characters mention a patch, hotfix, fix or update.
Limits: keyword matching counts “no bugs” as a mention; English reviews only; games released late in the month had fewer days of reviews before the scrape; silent depot updates without an announcement are missed; one month is one data point. The script caches every response, so an edition can be rerun exactly. Corrections go to contact@jumbojelly.com.
What should a studio do with this?
Three things, in order. Expect patch day: five in six launches patch within a month, so plan who pushes the fix and run the patch day QA checklist before it goes live. Test the first hour hardest: the bug mentions come from reviews written in the first 14 days, mostly by players who did not get far. And measure your own launch the same way: count the share of reviews mentioning bugs in your first two weeks and compare it with your genre’s median above. Edition 2, for October 2026 releases, comes in early November.