ScholarXIV parses through hundreds of research papers and reads multiple web pages before generating a response. Every answer is grounded in cited references rather than relying solely on the model's training data, making it suitable for serious academic work. You can access the referenced papers and webpages directly in the chat, so every claim is verifiable.

Yes. The Selection feature lets you click on multiple papers or highlight specific text inside them to build a custom context for your research. Selected papers and content are clearly indicated with an emerald green highlight on the right sidebar and above the chat input, so you always know what the AI is referencing.

The Sandbox is a built-in code execution environment that lets your research agent run code, simulations, calculations, and more — all within your research workflow. Instead of relying only on next-token prediction, your agent can write and run code to calculate results, test hypotheses, generate graphs and figures, analyze data, and export files, all in one place.

Pulse is a proactive feature that automatically generates easy-to-digest summaries and cross-research insights from your recently liked and bookmarked papers. There is no setup required — Pulse reads through your library and surfaces patterns, comparisons, and connections between papers, saving you from having to open each one manually. The more papers you save, the more valuable Pulse becomes.

Yes. Collections are shareable lists of papers that you can curate around any topic or project. You can invite others as editors or viewers, create shareable links, and optionally set your collection to public so it appears in the Discover Collections feed for the entire community to find.

Yes. Research outputs, summaries, and generated reports can be exported as PDF or Markdown, preserving the styling and formatting of your work. This makes it easy to save, share, or publish your findings outside of ScholarXIV.

ScholarXIV offers 100+ frontier, open-source, and latest AI models powered by Vercel's AI Gateway, ranging from advanced reasoning models to cutting-edge open-source options. Even models that don't natively support tool calls work seamlessly — we use a GPT-oss-20b fallback mechanism to handle tool execution, giving you full flexibility regardless of which model you choose.

Yes. The Free plan gives you full access to explore papers, use the AI chat, and test core features with generous usage limits. When you are ready for higher rate limits, file uploads, watermark-free exports, and priority support, you can upgrade to Plus or Pro at any time.