NewsBlur
The open-source RSS reader that learns what you like and filters the rest.
About
NewsBlur is an open-source RSS reader founded in 2009 by Samuel Clay, who has run it as a one-founder independent business ever since. Its headline idea is intelligence training: instead of just subscribing, you teach the reader to promote or hide stories by author, tag, title text, and keyword. Over time the river of news reshapes itself around what you actually read. The hosted service at newsblur.com runs on a limited free tier of 64 sites; Premium is $36/year for 1,024 sites, Premium Archive is $99/year for 4,096 sites with full-text training and web feeds (following sites without RSS), and Premium Pro at $29/month covers 10,000 sites with faster fetching.
Everything — backend Python, frontend JavaScript, and the native iOS and Android apps — is published on GitHub under the MIT license (about 7,700 stars), and you can self-host the whole stack with Docker. That makes NewsBlur unusual: the only open-source reader that is simultaneously a real hosted service and a self-hosted app. Native extras include newsletter forwarding, full-text search, a social layer of public blurblogs for shared stories, and an Ask AI feature that queries stories through Claude, GPT, Gemini, or Grok. Third-party clients like Reeder, ReadKit, and Unread work with it. Honest tradeoffs: the free tier's 64-site cap is the tightest of the big readers, unread items expire for free users, and the interface shows its age next to Feedly and Inoreader.
What it can do
- Intelligence training: promote/hide stories by author, tag, title, text, URL, or regex
- Hosted service (free 64 sites) or self-hosted via Docker — your choice
- Native iOS and Android apps; third-party clients via API (Reeder, ReadKit, Unread)
- Newsletter forwarding addresses pull email into your feeds
- Web feeds: follow sites without RSS (Premium Archive)
- Full-text search and full-article reading with story change tracking
- Public blurblogs: share stories to a personal blog and follow other curators
- Ask AI queries stories through Claude, GPT, Gemini, or Grok
- IFTTT integration, keyboard shortcuts, and a public API
Setup
Sign up at newsblur.com (free up to 64 sites) or pull the Docker image from GitHub to self-host. Import an OPML file or add feeds by URL; forward newsletters to your NewsBlur address. Train the intelligence by marking stories liked/disliked by author, tag, title, or keyword. Upgrade to Premium ($36/year) for 1,024 sites. Install the native iOS/Android apps or connect Reeder, ReadKit, or Unread via the API.
Prerequisites
A web browser or the iOS/Android app for the hosted service; Docker for self-hosting; Premium subscription for more than 64 sites.
Use cases
Works with
Trust score — 9/10
Scored 9/10 (verified tier, metrics-v1). Evidence: founded 2009 by Samuel Clay with Wikipedia-documented history (1,500 to 60,000 users in the 2013 Google Reader exodus); full stack MIT-licensed on GitHub with 7,662 stars and commits pushed October 16, 2026; published pricing tiers independently corroborated across 2026 reviews. Strengths: genuinely open (host or self-host), unique intelligence training, sustainable one-founder paid model, native mobile apps. Docked for: (1) free tier limited to 64 sites — tightest of the major readers; (2) the interface feels dated next to Feedly/Inoreader. License: MIT.
★ 7,662 GitHub stars
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 8.0/10
- Maintenance · recency, releases, issue load
- 9.0/10
- Openness · license + docs
- 10.0/10
- Manual review · sanity read + red-flag check
- 9.0/10
- · 7,662 GitHub stars (Oct 16, 2026 snapshot), commits pushed same day; MIT; founded 2009 by Samuel Clay (one-founder independent business)
- · Free 64 sites; Premium $36/yr (1,024 sites); Premium Archive $99/yr (4,096); Premium Pro $29/mo (10k); Docker self-host available
- · Only open-source reader that is both a real hosted service and self-hosted; native iOS/Android apps; Ask AI via Claude/GPT/Gemini/Grok
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.