Readest

There are a lot of good ideas here, but it suffers from the quantity-over-quality issue common to most heavily LLM-dependent projects (I am reminded of Dify, which does everything but crashes due to elementary parallelism issues in knowledge pipelines). The #2 contributor is "Claude Code" and tracks the primary contributor's activity (because it's attached as a co-author). This is typical of my experience speaking to developers using LLMs this heavily, and using the products of their and others such projects. As a software engineer, I can understand how this can be alluring to developers: oftentimes it can feel like one has writers block, or doesn't have the time to do everything that they could possibly want. For certain people, the result of combining code generation tools with those fundamental wants can result in projects that are just smorgasbords of stuff. For example, the website is fancy and all that, the README is full of extensive feature checklists, every release has an absolutely prodigious number of build artifacts including Debian and Redhat packages, AppImages, portable and installer binaries for Windows - all of these for amd64 and arm64 - as well as universal and arm64 android packages, and of course the single universal macOS package. Mind you this is all HTML/JS running in a platform web view with Tauri, so the architecture-dependent portion should be minuscule.

That long preamble aside, the user interface isn't very well suited to mobile devices, even with large screens: things like the font menu being a combobox that is clipped to the bounds of the settings modal. The settings modal should not be a modal on any mobile device, it should be a full screen view, and drop-down/up comboboxes should use selection modals on mobile. "Importing" fonts being the only means of loading fonts rather than allowing to search a defined director y (as koreader does) makes support for split fonts (i.e. Charis comes as "Charis-Regular," "Charis-Bold," and so forth) unclear and very tedious. Page layout settings could take a major cue from koreader's design and both appear not in the center of the screen blocking the page view (try the bottom) and could do to visually illustrate the effects of what it is they do (i.e. font size, etc...). Operating with "e-ink" mode should also change certain behaviors like dimming the page when showing the search box (even with our fancy modern instant-response eink displays). Online features such as which rely on translation or inference services don't appear to allow the user to bring their own API keys (one must build a custom image). The list could go on.

The gist of this is that the rough functionality is there. There are some really good ideas here, like word lens, but a lack of functional polish that is present in more standard packages like koreader.

The performance is horrendous, and it crashes constantly. Although I am not surprised, since it's an AI vibe-coded slop made with Claude by a chinese developer; this page claims it's US origin which is a blatant lie.

Author: x.com/hwangxin

Readest application is the best eBook reader app I have installed so far. Its features are powerful to upgrade my reading experience. I really appreciate the cloud synchronisation and application customisation.

Readest is a feature-rich, easy-to-use, free and open source ebook reader with no ads. It's available on every platform you could want it on, and is boasts great, modern UI design.

I've been using it for quite some time and am very satisfied. It's full of features and fast.

This one worked on my PostmarketOS phone when even Foliate struggled with the Text to Speech (TTS). Aside from that it is also a highly functional piece of software.