Get Ansel for Linux, Windows and macOS, see how stable the nightly builds are, and pick any recent build of every package with its crash rate and the number of people who ran it.

Install

Linux

Distribution-agnostic, portable AppImage executable Download latest

Windows

Windows 7 to 11 installer Download latest

MacOS X

Distribution for MacOS X 15 and newer. Download latest (Apple M)

Download latest (Intel)
Build from source

Best performance for your hardware

Building instructions

Installation help
Getting started

Is it stable ?

Stats updated automatically every 4h from opt-in Sentry.io crash logs collection. Last update :

The two charts on the top row read the whole fleet, the two below read one revision at a time. Wherever a date appears, it is the date the build was committed, never the day someone happened to be editing.

Crash-free sessions by build date : one line per operating system, each averaged over about a week. The dates are when the code was written, not when it was run — a crash today on last week’s nightly counts against last week’s build. So reading left to right shows the code getting better, not the calendar going by. A bad build dips at its own date, and the lines say which platform it hit.

Where crashes happen, and where bug reports come from : Windows carries just over half of all crashes — and crashes nearly three times more often per session than Linux — yet files under a quarter of the bug reports. Linux is the mirror image. macOS sits closer to Linux than its share of reports would suggest, and that is not luck : it is a UNIX, so a good deal of what we fix on Linux reaches it for free. Windows shares almost none of that code — it is a different architecture, not a variant — so a Windows bug exists until a Windows user tells us about it. A bug nobody reports is a bug nobody can fix, and it keeps crashing for everyone on that platform. Ansel is free, and written by people who cannot test on your machine : telling us what breaks is the contribution we actually need. Whatever happens, your edits are written to disk as you make them, so a crash costs you the last action at most.

Ansel has not published a stable release yet and there is no ETA for one : a new release is published when the list of all bugs have been cleared, so we know the software and stable. So far, Ansel only publishes revisions, which are intermediate states of the sourcecode. Once a revision has been tested by at least 25 unique users, its reliability stats can show above : the charts display the 30 most recent revisions that reached that threshold. An asterisk (*) marks revisions we shipped as a packaged nightly build.

On the two per-revision charts above, the crash-free rate is shown as a percentage whose bar opacity encodes our confidence : the solid part (almost certain) is the rate we are confident the software reaches, the likely range goes up to the currently recorded average, then optimistic reaches up to the best plausible rate written in parentheses on top. The fewer people tested a revision, the more incertainty there is on the average, the larger the {likely + optimistic} range gets : it is the margin of error of the current average at 95% confidence.

Bugs need to be reported on Github  or they will never be fixed. This free software is only guaranteed to work on the computers of its developers, who are not in front of yours. Ansel saves your editing histories immediately after each change: in case of a crash you loose at most the last action.

The links above always point to the latest nightly build of the “fairly stable” branch. If you want a particular revision or need to roll back, you can find all intermediate versions on Github .

Every recent build, per package

Builds are named Ansel-x.y.z+N.g<commit>: a higher N is a newer build, and the commit links to the exact source it was built from. Nightlies are the “fairly stable” channel — quickly broken, quickly fixed — so the crash-free rate is there to help you pick: a build that many people ran without crashing is a safer bet than yesterday’s. Reading the columns:

  • Crash-free — share of sessions on that build, on this platform, that ended without a crash. Unmarked: measured by the opt-in crash reports (Sentry) filed from this platform. ≈: estimated the way the chart above is — crash reports from this platform (Sentry) over sessions started on it (opt-in usage statistics, PostHog), two populations stitched together. *: the build’s rate over every platform, nothing platform-specific being available. The two macOS packages share one figure, as no architecture is reported.
  • Testers — how many distinct installations ran that build on this platform: the opt-in usage statistics (PostHog) and the opt-in crash reports (Sentry) merged on the anonymous install id both carry, so an installation seen by either counts once — the same merge as the unique-users chart. Hover for the two sources. The two macOS packages share one count.

Linux AppImage

BuildDateCommitSizeCrash-freeTesters
0.0.0+5241.g7230d1b355 2026-10-087230d1b355 252 MiB100.0 % ≈14
0.0.0+5232.gea59057346 2026-10-06ea59057346 252 MiB96.5 % ≈16
0.0.0+5214.g0dd4030353 2026-10-050dd4030353 252 MiB100.0 % ≈7
0.0.0+5203.g8bef6908d9 2026-10-048bef6908d9 252 MiB100.0 %14
0.0.0+5201.g65e7a3adea 2026-10-0265e7a3adea 252 MiB98.9 %19
0.0.0+5192.ga4eac9fe11 2026-10-01a4eac9fe11 252 MiB100.0 %27
0.0.0+5173.g73980c3cf0 2026-09-3073980c3cf0 252 MiB100.0 %15
0.0.0+5141.gf37105c227 2026-09-29f37105c227 252 MiB100.0 % ≈11
0.0.0+5117.g4e7fd541ce 2026-09-274e7fd541ce 252 MiB100.0 %15
0.0.0+5112.ge83ba5865d 2026-09-26e83ba5865d 252 MiB100.0 % ≈13
0.0.0+5111.gced9bb3faa 2026-09-26ced9bb3faa 252 MiB97.8 %17
0.0.0+5102.g4c9de250d8 2026-09-244c9de250d8 252 MiB99.2 %37
0.0.0+5090.g88cb31d572 2026-09-2388cb31d572 251 MiB98.8 %23
0.0.0+5085.g4c10c57faf 2026-09-224c10c57faf 252 MiB100.0 % ≈8
0.0.0+5038.gf484018567 2026-09-17f484018567 251 MiB96.4 %44

Linux Flatpak

BuildDateCommitSizeCrash-freeTesters
0.0.0+5241.g7230d1b355 2026-10-087230d1b355 98 MiB100.0 % ≈14
0.0.0+5232.gea59057346 2026-10-06ea59057346 99 MiB96.5 % ≈16
0.0.0+5214.g0dd4030353 2026-10-050dd4030353 98 MiB100.0 % ≈7
0.0.0+5203.g8bef6908d9 2026-10-048bef6908d9 98 MiB100.0 %14
0.0.0+5201.g65e7a3adea 2026-10-0265e7a3adea 99 MiB98.9 %19
0.0.0+5192.ga4eac9fe11 2026-10-01a4eac9fe11 99 MiB100.0 %27
0.0.0+5173.g73980c3cf0 2026-09-3073980c3cf0 99 MiB100.0 %15
0.0.0+5141.gf37105c227 2026-09-29f37105c227 99 MiB100.0 % ≈11
0.0.0+5117.g4e7fd541ce 2026-09-274e7fd541ce 98 MiB100.0 %15
0.0.0+5112.ge83ba5865d 2026-09-26e83ba5865d 99 MiB100.0 % ≈13
0.0.0+5111.gced9bb3faa 2026-09-26ced9bb3faa 98 MiB97.8 %17
0.0.0+5105.gf7549424c2 2026-09-24f7549424c2 99 MiB95.7 % ≈6
0.0.0+5102.g4c9de250d8 2026-09-244c9de250d8 99 MiB99.2 %37
0.0.0+5090.g88cb31d572 2026-09-2388cb31d572 100 MiB98.8 %23
0.0.0+5086.g99666b0d68 2026-09-2299666b0d68 99 MiB92.9 % ≈6

Windows installer

BuildDateCommitSizeCrash-freeTesters
0.0.0+5241.g7230d1b355 2026-10-087230d1b355 147 MiB100.0 % ≈14
0.0.0+5203.g8bef6908d9 2026-10-048bef6908d9 147 MiB97.4 % ≈16
0.0.0+5201.g65e7a3adea 2026-10-0265e7a3adea 147 MiB98.6 % ≈18
0.0.0+5192.ga4eac9fe11 2026-10-01a4eac9fe11 147 MiB95.5 % ≈15
0.0.0+5173.g73980c3cf0 2026-09-3073980c3cf0 147 MiB100.0 %14
0.0.0+5141.gf37105c227 2026-09-29f37105c227 147 MiB95.2 % ≈8
0.0.0+5117.g4e7fd541ce 2026-09-274e7fd541ce 147 MiB100.0 % ≈13
0.0.0+5112.ge83ba5865d 2026-09-26e83ba5865d 147 MiB97.5 %20
0.0.0+5111.gced9bb3faa 2026-09-26ced9bb3faa 147 MiB100.0 % ≈20
0.0.0+5105.gf7549424c2 2026-09-24f7549424c2 147 MiB100.0 % ≈14
0.0.0+5090.g88cb31d572 2026-09-2388cb31d572 147 MiB100.0 % ≈13
0.0.0+5089.gaa1090dd07 2026-09-23aa1090dd07 147 MiB100.0 % ≈4
0.0.0+5085.g4c10c57faf 2026-09-224c10c57faf 147 MiB100.0 % ≈1
0.0.0+5038.gf484018567 2026-09-17f484018567 147 MiB99.1 %29
0.0.0+5020.gb35c95990d 2026-09-16b35c95990d 147 MiB100.0 %10

macOS, Apple Silicon

BuildDateCommitSizeCrash-freeTesters
0.0.0+5241.g7230d1b355 2026-10-087230d1b355 140 MiB100.0 % ≈4
0.0.0+5232.gea59057346 2026-10-06ea59057346 140 MiB83.3 % ≈4
0.0.0+5214.g0dd4030353 2026-10-050dd4030353 140 MiB100.0 % ≈5
0.0.0+5203.g8bef6908d9 2026-10-048bef6908d9 140 MiB75.0 % ≈4
0.0.0+5201.g65e7a3adea 2026-10-0265e7a3adea 140 MiB85.7 % ≈10
0.0.0+5192.ga4eac9fe11 2026-10-01a4eac9fe11 140 MiB90.0 % ≈6
0.0.0+5173.g73980c3cf0 2026-09-3073980c3cf0 140 MiB100.0 % ≈3
0.0.0+5141.gf37105c227 2026-09-29f37105c227 140 MiB100.0 % ≈4
0.0.0+5117.g4e7fd541ce 2026-09-274e7fd541ce 140 MiB63.6 % ≈4
0.0.0+5112.ge83ba5865d 2026-09-26e83ba5865d 140 MiB90.9 % ≈4
0.0.0+5111.gced9bb3faa 2026-09-26ced9bb3faa 140 MiB100.0 %7
0.0.0+5105.gf7549424c2 2026-09-24f7549424c2 140 MiB100.0 % ≈4
0.0.0+5090.g88cb31d572 2026-09-2388cb31d572 139 MiB100.0 % ≈4
0.0.0+5085.g4c10c57faf 2026-09-224c10c57faf 139 MiB83.3 % ≈2
0.0.0+5038.gf484018567 2026-09-17f484018567 139 MiB100.0 %12

macOS, Intel

BuildDateCommitSizeCrash-freeTesters
0.0.0+5232.gea59057346 2026-10-06ea59057346 142 MiB83.3 % ≈4
0.0.0+5203.g8bef6908d9 2026-10-048bef6908d9 142 MiB75.0 % ≈4
0.0.0+5201.g65e7a3adea 2026-10-0265e7a3adea 142 MiB85.7 % ≈10
0.0.0+5173.g73980c3cf0 2026-09-3073980c3cf0 142 MiB100.0 % ≈3
0.0.0+5141.gf37105c227 2026-09-29f37105c227 142 MiB100.0 % ≈4
0.0.0+5117.g4e7fd541ce 2026-09-274e7fd541ce 142 MiB63.6 % ≈4
0.0.0+5112.ge83ba5865d 2026-09-26e83ba5865d 142 MiB90.9 % ≈4
0.0.0+5111.gced9bb3faa 2026-09-26ced9bb3faa 142 MiB100.0 %7
0.0.0+5105.gf7549424c2 2026-09-24f7549424c2 142 MiB100.0 % ≈4
0.0.0+5102.g4c9de250d8 2026-09-244c9de250d8 142 MiB99.3 % *—
0.0.0+5090.g88cb31d572 2026-09-2388cb31d572 142 MiB100.0 % ≈4
0.0.0+5085.g4c10c57faf 2026-09-224c10c57faf 142 MiB83.3 % ≈2
0.0.0+5038.gf484018567 2026-09-17f484018567 142 MiB100.0 %12
0.0.0+5020.gb35c95990d 2026-09-16b35c95990d 142 MiB100.0 % ≈4
0.0.0+5018.ga010adc904 2026-09-15a010adc904 142 MiB100.0 % ≈2

Docker image

BuildDateCommitSizeCrash-freeTesters
0.0.0+5241.g7230d1b355 2026-10-087230d1b355 768 MiB100.0 % ≈14
0.0.0+5232.gea59057346 2026-10-06ea59057346 768 MiB96.5 % ≈16
0.0.0+5214.g0dd4030353 2026-10-050dd4030353 769 MiB100.0 % ≈7
0.0.0+5203.g8bef6908d9 2026-10-048bef6908d9 769 MiB100.0 %14
0.0.0+5201.g65e7a3adea 2026-10-0265e7a3adea 768 MiB98.9 %19
0.0.0+5192.ga4eac9fe11 2026-10-01a4eac9fe11 768 MiB100.0 %27
0.0.0+5173.g73980c3cf0 2026-09-3073980c3cf0 770 MiB100.0 %15
0.0.0+5141.gf37105c227 2026-09-29f37105c227 768 MiB100.0 % ≈11
0.0.0+5117.g4e7fd541ce 2026-09-274e7fd541ce 768 MiB100.0 %15
0.0.0+5112.ge83ba5865d 2026-09-26e83ba5865d 768 MiB100.0 % ≈13
0.0.0+5111.gced9bb3faa 2026-09-26ced9bb3faa 768 MiB97.8 %17
0.0.0+5105.gf7549424c2 2026-09-24f7549424c2 768 MiB95.7 % ≈6
0.0.0+5102.g4c9de250d8 2026-09-244c9de250d8 768 MiB99.2 %37
0.0.0+5090.g88cb31d572 2026-09-2388cb31d572 768 MiB98.8 %23
0.0.0+5085.g4c10c57faf 2026-09-224c10c57faf 768 MiB100.0 % ≈8

Crash-free rate and testers cover the last 90 days of opt-in reports; a dash means no report reached us for that build.

Translated from English by : Automatically generated. In case of conflict, inconsistency or error, the English version shall prevail.