Thrilled to be named to @FortuneMagazine's 2026 #Future120! 🚀 For the 3rd year, Fortune recognized Cloudflare's long-term growth and innovation. Helping to build a better Internet is a long-term commitment and we’re grateful to be recognized along the way. fortune.com/ranking/future/?…

I am of the strong belief that a large part of the future internet economy collapses into Agents or Services. MCP will be how they communicate. Both will be mostly headless.

GPT 6.1 Sol in Fast mode can now hit 100+ Tokens per second. Given how token efficient the model is, it feels faster than Opus, and it's not outrageously expensive like Ultrafast. This makes codex much more relevant to my work again. Thanks @thsottiaux.

Introducing HyperFrames Studio a video editor built for agents, from scratch Describe it, your agent builds, then you both work on the same timeline Draw or comment on a frame, and it makes the edit 100s of inspirations & templates Download ↓

Why still use Codex if Claude Opus/Sonnet 5.5 are better and faster Just trying to understand

pewdiepie’s character development needs to be studied. > dropped out of university. > worked odd jobs while uploading gaming videos. > became the biggest YouTuber. > got into reading books. > got into drawings (insanely good) > got into philosophy. > moved to Japan and started family vlogs. > got into tech. > deep into linux. > installed arch. > replaced all his google services. > learned coding. > bought tons of GPUs. > made his own local data center. > trained his own LLM. > distilled OpenAI’s Sol into his local model and got banned twice lmao.

Hace unos meses, PewDiePie, que fue el youtuber más grande del mundo durante casi una década con videos de juegos y sin nada que ver con tech, se puso a vibecodear y armó Odysseus: su propio harness de IA, local y open source. Y hoy lanza Ajax, un modelo de IA basado en Qwen 3.5 9B y afinado para ese harness. La idea: darle a tu agente un modelo que vive en tu PC y resuelve lo cotidiano (buscar, navegar, gestionar mails y calendario) sin mandar nada a un datacenter. Para armar los datos de entrenamiento intentó destilar Sol de OpenAI, pero le banearon la cuenta dos veces. Aún no está disponible. Sale el 3 de octubre a las 08:25 (hora de Japón). Más info: data.pewdiepie.com

Never ever using @framer again in my life We exceeded the bandwidth allowed one month, we took action to fix it but it turns out we didn't fix the biggest source of the issue and exceeded another consecutive month....fair enough, we should pay for the extra bandwidth we consumed, more than happy to do that. However, forcing me to buy 3x additional bandwidth on a yearly basis to access my site ($1621) and publish edits is the most anti creator concept. Enterprises, sure, def not for creators. Just because my bandwidth spiked 1-2 months doesnt mean its gonna continue that way for the next 12 months. Just charge me for what I overused. Always loved framer, still do, and spent hours using their products and loved every second of it. This is not a hate tweet, its just my honest opinion

Eleven v4 is INSANE! 🤯 Here's my 44-sec spec ad for it. I wrote a script of everything an AI voice "can't do." Then made Eleven v4 read it out loud. Sound on 🔊

Claude Opus 5.5 just took a big drop on NerfBench. Yesterday it was scoring above launch. Today it's at 94.2%. GPT 6 Astra: 98.0% Sonnet 5.5: 100.9% GPT 6.1 Sol: 106.7% 94.2% is still inside normal variance, so we can't call it a nerf yet. But we're watching Opus 5.5 very closely.

Hace algunos meses una conocida me dijo que le iban a hacer la nueva versión de su web en framer. La tenía en Wordpress y un familiar suyo diseñador se la iba a diseñar y hacer en framer. Le comenté que no creo que fuera muy buena idea, pero bueno... Hoy me viene que se han cargado todo el SEO y que si le puedo ayudar (error 404 everywhere) Me meto en framer para hacer 301 y... me he encontrado el precio de la facilidad 😂

Never complaining about developing for Shopify ever again... This was brutal 🙄

Agents already speak git. Artifacts lets you create and fork repos from a Worker, hand out tokens, and clone with any git client, with events on every push. Now in open beta on the Workers Paid plan. Build the next GitHub? blog.cloudflare.com/next-git… Here's how it works:

excited to announce that @Cloudflare Artifacts is now in open beta! with Artifacts, you can give every application and agent its own repo that you can manage programmatically plus you can use it to store and deploy your Workers code (and ofc it integrates with Workers Previews)

This is the new expectation level for Shopify apps: You need to be lower cost than vibecoding, or you won't be viable as a business. Enterprise Saas is beyond fucked.