The UnifiedToolchain for the Webdev, build, test, lint, format, monorepo caching & more in a single dependency, built for scale, speed, and sanityEverything you love about Vite— plus everything you’ve...
Announcing Vite+: a unified toolchain for JavaScript Last week, we unveiled Vite+ at the first-ever in-person ViteConf in Amsterdam. In this post, we’ll share more details about what it is and the motivation...
Cloudflare investigated CPU performance benchmark results for Workers, uncovering and fixing issues in infrastructure, V8 garbage collection, and OpenNext optimizations. These improvements have made Cloudflare Workers faster for all customers.
URL Snake Play the classic snake game on a URL! https://demian.ferrei.ro/snake This is how the game should look: Note that the game might be unplayable on some browsers for different reasons, like the...
Introducing Claude Haiku 4.5 (via) Anthropic released Claude Haiku 4.5 today, the cheapest member of the Claude 4.5 family that started with Sonnet 4.5 a couple of weeks ago. It's priced at $1/million input...
Claude Haiku 4.5, our latest small model, is available today to all users.What was recently at the frontier is now cheaper and faster. Five months ago, Claude Sonnet 4 was a state-of-the-art model. Today,...
In a joint study with the UK AI Security Institute and the Alan Turing Institute, we found that as few as 250 malicious documents can produce a "backdoor" vulnerability in a large language model—regardless of...
Is Lovable dying? Web traffic has declined almost 50% from 35.4M in June to 19.1M in September. This trend can be seen across other big vibe coding tools such as Replit and Bolt. Here’s why I think this is happening: 1) The viral wave has passed: Spring/summer hype (esp. Lovable’s June peak) pulled in lots of non-devs on monthly plans. They experimented with the tool and never returned.The cultural conversation has also shifted from “AI can build an entire app” back to a more realistic view of AI’s limitations. 2) Power users eating up compute: A small group of power users burned through compute budgets. To protect margins, platforms throttled usage and raised prices. That scared away hobbyists and experimenters, reducing the casual exploration that inflated spring traffic. 3) Lovable’s need to monetize: After raising $200M, Lovable raced to monetize, tightening free credit limits and ending generous 50% discounts. The friction dulled its appeal for casual builders. 4) In May, v0 lost 500k (10%) of web traffic after moving from $20/month to usage-based credits. Casual users found their balance disappearing in hours, leading to outrage. 5) Replit’s missteps: Replit once thrived on its free tier and always-on Repls. But over the years, free education plans discontinued, always-on removed from basic plans. A high-profile AI agent failure occurred in July (accidentally deleting a live database) and user trust took a hit. 6) AI coding losing its novelty: AI coding is being absorbed into mainstream tools. GitHub Copilot, Cursor, and VS Code integrations keep developers in their IDEs, instead of web-first “vibe coding” sites. Google’s Firebase Studio also launched AI-powered app building with Gemini inside Google Cloud, eating into the traffic of these sites. However, new entrants like Emergent have managed to keep growing despite bigger players losing traffic. It could be that the tool is still in its hype phase - YC acceptance, fresh funding has helped it scale. The real test will be whether it avoids the same hype to drop-off curve. Vercel’s v0 has also seen an increase in traffic likely off the back of its new Marketplace and deeper Vercel integrations. And ecosystem pull matters, teams already shipping on Vercel/Next.js can try v0 by default. They also raised funding at the end of September, which could have resulted in an increase in web traffic. It will be interesting to see how players like Lovable, valued at a billion dollars, will maintain growth and valuation. | 326 comments on LinkedIn
Bun 1.3 introduces zero-config frontend development, unified SQL API, built-in Redis client, security enhancements, package catalogs, async stack traces, VS Code test integration, and Node.js compatibility improvements.
The Story of Codesmith: How a Competitor Crippled a $23.5M Bootcamp By Becoming a Reddit Moderator
(larslofgren.com)Let’s say you decide to start a coding bootcamp. Your background is in pedagogy and you love teaching. Your parents were teachers. You find a co-founder, raise a bit of money, and pour your soul into your company. The first couple of years, students love your program. Positive feedback, extraordinary student outcomes, employees love the mission. You are quite literally […]
Microsoft is plugging more holes that let you use Windows 11 without an online account
(www.theverge.com)Tom Warren is a senior editor and author of Notepad, who has been covering all things Microsoft, PC, and tech for over 20 years.Microsoft is cracking down on bypass methods that let Windows 11 installs use a...
7 min readSep 30, 2025--It’s here! Long have the rumours been kept not-super-secret. Vega, the new TV (and Echo box) OS from Amazon is finally out in the open. And the rumours around React Native were true,...
I first wrote about AV1 in April 2019 when it was just a baby codec. I covered the monthly ecosytem updates a bunch more times (which you can find in the archives) eventually culminating in the AV1 2019: A...
7 min readSep 17, 2025--We Love You, TypeScriptFor nearly five years now, Motion has operated in a large TypeScript monorepo. At its peak, it was roughly ~2.5 million lines of code after excluding comments,...
It turns out we've all been using MCP wrong. Most agents today use MCP by exposing the "tools" directly to the LLM. We tried something different: Convert the MCP tools into a TypeScript API, and then ask an LLM to write code that calls that API. The results are striking.
Exploring how React's dominance by default stifles frontend innovation, and why deliberate framework choices lead to better tools for performance, developer experience, and ecosystem diversity.
CloudFlare vs Vercel benchmark This repo is meant to benchmark SSR performance between CloudFlare and Vercel. There's a lot of misinformation going around and I'm annoyed. So I wrote a really annoying...
Viewing most recent comments 1 to 40 of 78 · Next page · Last pageBoss level done! I consider this as beating the game. It breaks after this anyway :D Cannon tower OP! As you can see, both this image and...
Admin Supporter Edit page Go to dashboard Admin ...
Accompany Armin, a young apprentice, on the swashbuckling adventure of a lifetime to discover the fate of his lost father, learn the magical arts of blacksmithing, and rise to meet a destructive force threatening his homeland before time runs out.
Performance cookies and advertising cookies Performance cookies These cookies help to improve the performance of BenQ. If you want to opt-out of advertising...
Discover ...