**Listen to papers: new AI audio overview** Commutes, queues, workouts — time you couldn't spend on papers is now listening time. PaperFeed generates a narrated overview that walks through a paper's problem and approach; play at any speed with line-by-line captions. Takes a few minutes to generate, and you'll get a push when it's ready. One per day free, unlimited on Pro. **Paper page: clearer entry points** AI summary, audio, and explainer video now sit in one place; repos, models and project pages mentioned in the abstract turn into tappable links. **Trending & recommended: pick a time window** Filter by the last 24 hours, 7 days, or 30 days, so older papers stop crowding the top. **Your reading, exportable** Copy BibTeX from any paper; export your library as Markdown with every highlight and note included. **Fixes** Pro monthly quota now resets correctly in all time zones; conference reviews handle the new ICLR 2025+ review form; pushes land on the paper page.

**A reworked paper page** One clean page per paper: abstract and quick actions first, code, reviews and citation networks grouped into Explore and Impact. Feed, search or citations — every way in now shows the same complete page. **AI explainer videos** When reading stalls, listen instead: generate a narrated video that explains the paper's method and contribution in your language. Ready in 10–20 minutes; you'll get a push when it's done. (Pro) **Faster and steadier** Polish across translation, AI chat, and reading.

**PaperFeed in your language** The app now speaks Traditional Chinese, French, Japanese, Korean, and Portuguese (Brazil) — in the app and on its App Store page. **Rate PaperFeed in one tap** Enjoying a paper? Rate us with a single tap — we'll only ask once you've had time to enjoy the app. **Polish & fixes** Smoother reading, a more responsive AI chat, and general stability improvements.

Bilingual reading, built for deep dives A new "Bilingual" mode in the reader shows the full paper paragraph by paragraph, original and translation side by side. The first translation takes ~20 seconds — after that it opens instantly. Free plan previews the first 8 paragraphs of every paper; Pro unlocks the whole thing. Try Pro free Subscriptions now start with a free trial with every Pro feature unlocked. Cancel at least 24 hours before it ends and pay nothing. Share to get +1 AI query Share PaperFeed with a classmate or colleague and get +1 AI query for the day — once per day. Faster, more reliable translation & summaries Title, abstract, and full-text translation run on an upgraded engine for better quality and stability; summaries of papers you've already opened appear instantly; brand-new papers become readable sooner. Long conversations in Chinese now keep their context intact.

Brand-new built-in reader Read full arXiv papers right inside the app — no more jumping to Safari. Clean layout, adjustable font size, dark mode, and automatic resume from where you left off. Ask AI about any passage Select any paragraph, equation, or term while reading and ask AI instantly — answers come with the full context. Peer reviews from top conferences Powered by OpenReview: read ICLR, NeurIPS and more right on the paper page — decision, average review scores, strengths/weaknesses/questions from each reviewer, and the Meta Review. Two new ways to discover papers · Trending: the daily Hugging Face hot list — see which papers the community is upvoting and starring · For You: recommendations that keep learning from your library — the more you save, the better it gets Code & ecosystem at a glance A new card on the paper page shows the official code repo with stars and framework, plus related models, datasets, and live demos — go from reading to reproducing. Daily paper digest Each morning, get a curated selection of new papers worth your time. Tap a notification to jump straight to the paper. Search & reading polish Brand-new papers are searchable immediately; wide tables and long equations scroll horizontally; TikZ vector figures scale to fit.

• Better reading experience: LaTeX notation in abstracts is now converted to readable text • Sharing improvements: fixed WeChat compatibility, and shared links now show an instant preview • More reliable AI chat: fixed streaming replies dropping out, with more robust model handling • Fixed AI summary headings appearing as a flat list • Feeds now refresh automatically when your connection recovers • Improved data coverage for the top-cited papers ranking

AI Paper Assistant — Never read a paper alone again. 7 AI models at your fingertips Qwen 3.5, DeepSeek V4, Kimi K2.6, GLM 5.1, MiniMax M2.7 and more. Free users get access to base models; Pro unlocks all. Model list updates from the cloud — no app update required. Full-text understanding AI reads the entire PDF, not just the abstract. Every answer cites the exact section it draws from, so you can verify in seconds. One-tap deep reading Open any paper to get a structured AI report: research question, methodology, key findings — all at a glance. Top Cited Feed A new "Top Cited" mode surfaces the most influential papers in your fields, with citation counts right on every card. DeepL academic translation Translate titles and abstracts instantly. Switch between Original, Translated, and Bilingual side-by-side reading modes. Feed titles auto-translate for faster browsing. Explore the research lineage Citations, references, and similar paper recommendations — navigate the academic network with a single tap. Beautiful share cards Generate stunning web links for papers and AI conversations, with field-aware color themes and instant social media previews. Design overhaul Glass-morphism navigation bar, choreographed entrance animations, read-state dimming, skeleton loading, and full dark mode.

Share Papers & Conversations Generate beautiful share cards for any paper, or share your AI chat sessions as a public link — perfect for collaborating with colleagues. Top Cited Feed A new "Top Cited" feed mode surfaces the most-cited papers in your fields, powered by Semantic Scholar data. Feynman Deep Reading A new deep-reading mode breaks down complex papers using Feynman-style explanations, helping you truly understand the "why" behind the methods. Improved AI Chat Reasoning traces are now visualized in real time. Chat history syncs intelligently across sessions. Markdown rendering upgraded with full support for tables, diagrams, and code blocks. Upgraded AI Models Built-in models updated to the latest versions; Stability & Performance Fixed citation count display, improved caching for top-cited data, and optimized cold-start loading.

More AI Quota for Everyone Your AI usage limit has been significantly increased — ask more questions, chat with more papers, and explore deeper than ever before. AI Paper Chat — Talk to Any Paper Ask questions, get instant answers. Our AI-powered chat understands the full context of any paper, recommends smart follow-up questions, and even shows its reasoning process. Your conversation history is saved automatically. Instant Title Translation Every paper now shows a translated title in your language. Browse and discover research in a way that feels natural — no translation tools needed. Feed Search & Faster Browsing Find papers instantly with keyword search right in your feed. Enjoy smoother loading with refreshed skeleton screens and lazy-loaded citations and references. More Languages, More Users PaperFeed is now available in German and Spanish — joining English, Chinese, Japanese, and Korean. Polished Profile & Settings A redesigned profile page, clearer AI usage display, email update support, and a brand-new settings page. Bug Fixes & Improvements Improved network reliability, fixed text rendering issues in PDFs and chat, and corrected localization formatting.