Mayank Singh

Check out researchwith.ai. It’s your constant AI research companion. Read any PDF with the AI as your partner. ✍️Highlight and annotate your reading 🤖Ask powerful AI models questions 🗂️Organize your reading into folders 🌐Find new papers via conversation search

Makes my day

Just added a topics page on researchwith.ai An easier layout to view all the Trending and your Continue Reading papers in a vertical scroll with additional details right on the page to quickly scan through abstracts as well. Experience added after feedback from the community. Let us know what you think. Platform walkthrough in the comments

Inception Mercury 2 blog post: inceptionlabs.ai/blog/introduci…

New models just landed on researchwith.ai: 1. Grok 4.20 (pro tier) 2. GLM 5 Turbo (free tier) 3. Bytedance - Seed 2.0 Mini (free tier) 4. Bytedance - Seed 2.0 Lite (free tier) 5. Inception - Mercury 2 (free tier) Grok 4.20 is extremely fast, but I would like to call out a the last model here: Mercury 2 by the team at Inception. A first of its kind Reasoning Diffusion LLM. The speed is honestly mind blowing. If you're used to models streaming tokens one at a time, this one will feel different. The video below isn't sped up. I'm extremely excited to experiment with this. I already have a few ideas about how it might be uniquely suited to some of the tooling on the platform. I've shared a link to their announcement blog post in the comments. Definitely recommend giving it a read.

Introducing IDE style progress bar to show you status of the current paper. For new papers on the platform it takes us some time to index them and create metadata for accurate answers. If users asked the AI a question during these few seconds you'd get a warning on the UI telling you that the "Paper is still being processed" but you had no idea how long it'll take or if it is processed. This improves the experience and provides an update when the paper is ready. Let me know if you like the experience. Try here: researchwith.ai

GPT-5.4 just landed on researchwith.ai, and we're giving away 1 FREE month of Pro. Sign up now (no credit card required) to get access to: • Frontier AI models • Higher usage limits • Larger upload limits Recent updates to platform: 1. GPT 5.4 and GPT 5.4 Pro now available to Pro users (sign up now to access them for free!) 2. Papers now show Like👍 and View👀 counts from the community. 3. "Continue reading" carousel: Pick up right where you left off. The reader automatically scrolls to the last page you were on. 4. Custom Tags: You can now tag papers with custom labels in addition to organizing them in folders. 5. Improved Natural Language Search Previously our search was powered by the Perplexity API, but we weren't happy with the accuracy. So we built our own NL search system powered by an internal search index (currently covering arXiv papers). We're working on a blog post comparing our results vs other platforms, stay tuned. We would love your feedback. More than half the features on the platform came directly from community requests, and we're continuing to build the product with our users. Feel free to: • Comment here • DM me • Join our Discord (link in comments)

Added a hotly requested UX feature on researchwith.ai: Continue reading carousel on the homepage Quickly get back to the paper you were reading right. Auto scrolls to the page you were last at.

🖱️Clickable arXiv References in the PDF/HTML view now on researchwith.ai arXiv IDs in paper references are now automatically detected and made clickable - even when the original PDF has no embedded links. Supported in both the PDF view and HTML view. Click any reference just by hovering over the arxiv id text to open the paper directly on researchwith.ai in another tab. Happy researching!

Introducing inline references on researchwith.ai The AI responses now add inline references for the exact content used to generate a certain part of an answer, improving the grounding of their response. You can hover over the reference to quickly glance at it or click "Go to Page" to read the section in depth. Try it out and leave your feedback

Link to the book: incompleteideas.net/book/RLbook202…

Richard Sutton and Andrew Barto's RL book is a goldmine for anyone trying to understand Reinforcement learning in an iterative fashion from first principles to modern ideas. But it's also a tough read, at least for me. This is the perfect use case for researchwith.ai. It's not just limited to research papers. It's for any serious literature you want to think through properly. Today we increased our free tier to support PDFs up to 20MB, and Pro now supports up to 100MB. Here's me uploading a 43MB book: - Navigating via index - Highlighting key sections - Asking questions about specific passages Read at your own pace. With an infinitely patient companion. Link to the book in the comments.

@aakashgupta Love NotebookLM, but it’s missing a few features. What if you could do all that and more with any AI model of you choosing? Actually read the text. Not vibe read it. The AI is there to help you when you need it. Try: researchwith.ai

NotebookLM is the most underrated AI tool Google has made. I used one notebook to go from zero knowledge to a working prototype, customer discovery materials, and a product deck in a single afternoon. Here's your complete guide: news.aakashg.com/p/complete-gui…

@tankots @WisprFlow @WisprFlow sick!

We offered 5 people a Porsche 911 GT3 RS if they could get @WisprFlow to make a mistake It's the fastest and most accurate AI voice dictation app that's 3x more accurate than ChatGPT, Claude, or Siri. Today, we’re finally launching on Android. Download now: play.google.com/store/apps/det… As a part of the launch, we’re giving away 6 months of Wispr Flow Pro for free. Like, retweet and comment ‘Wispr Flow’ to get it. Enjoy. — Written with Wispr Flow

Exciting new update to researchwith.ai We've improved the context retrieval flow for our AI chat. Includes: 1. Hybrid search - Improved semantic matching (improved embeddings) + introducing keyword matching 2. Better visual context - Your AI companion always knows what you're currently reading (Current page context) Sign up to researchwith.ai today. We have an amazing free tier. Enter an arXiv id, PDF url, or upload your own PDF (only you have access)

One of the @claude code features that I feel hasn't received enough attention is the /insights command. Been using it for a couple of weeks and every few days I generate a report showing me a summary of my recent claude code sessions. I've found many helpful tips to speed up development and reduce errors, even provides the CLAUDE.md instructions and hooks to add based on failures

Say hello to new models on researchwith.ai! Now you can use: 1. Qwen 3.5 (free tier!) 2. Qwen 3.5 Plus (free tier!) 3. Claude Sonnet 4.6 (Pro tier) Qwen 3.5 now introduces reasoning with amazing performance for the price. Find, read, annotate and understand latest research publications on researchwith.ai! We have a very generous free tier and support more than 10 amazing models and more on the pro tier!