Paper Notes

Paper Notes hits a real sweet spot: it's free, enormous, and surprisingly well-organized. With over 23,000 notes, you can practically sample the entire output of 11 leading AI conferences. The breakdown by conference and 55+ subfields is practical—you can drill into, say, all 751 3D vision papers in CVPR2026 or 241 LLM reasoning papers in ICLR2026. That granularity makes it a genuine time-saver for literature surveys. We'd reach for this when we need to quickly answer 'what's hot in LLM agent papers this cycle?' or compare topic volume across conferences. It's also handy for newcomers trying to map out subfields. The lack of registration is a plus—no signup friction. Where it bites: the notes are AI-generated, so they're summaries, not critical reviews. Quality varies—some are crisp, others generic. There's no integrated PDF search or citation export, and no API for programmatic access. If you need deep scrutiny, author implementations, or structured data for analytics, this won't cut it. Compared to Papers With Code, Paper Notes is faster for browsing but offers no code links, benchmarks, or leaderboards. It's also more conference-specific than arXiv. Think of it as a complement to those resources—a quick-fix digest for staying current. In practice, we'd pair it with a proper paper search for the top picks. It's not a research tool, but a discovery and orientation tool. For that job, it's excellent and free.