Dator @drdator

@chewymouse The game is available here for anyone to play dator.dev/city2026/

Low-poly SimCity 2000 I (using Opus 5.5) built over a few days. I created it because I wanted it and to see how capable Opus 5.5 was. It turned out much better than expected. dator.dev/city2026/ @claudeai

@theo fucking idiots we were

@mattshumer_ Looks incredible

Holy crap, these numbers look good. I haven't had a chance to test the model yet... but if it's as good as these benchmarks say, it will be a massive step up from Fable 5.1. Just think about that... 5.1 is now outdated. This is what acceleration looks like.

Claude @claudeai

a month ago

Opus 5.5 is a major step up from Opus 5, leading on agentic coding, computer use, and knowledge work.

161 621 8K 1.6M 819

Play chess against @typesafeai's Jev jev-chess-eta.vercel.app

My three year old loves playing papaya drive dator.dev/papaya/ built with Astra, three.js and blender

If you don't constantly lose track of what your agents are doing, you’re not shipping hard enough.

I asked Astra for a low-poly car in Blender. Literally 3 more prompts and I had a playable game. Models, physics, gameplay. Astra is truly next level. d2e17ltpesncil.cloudfront.net

Release soon, I hope

Oh, this is the ghostty rendering the image ;)

Released q3edit.com today and it reached the the front page of HN(!).

Fable 5: Doom Bench™ completed. It’s by far the best one so far - except that the gun look like a set of testicles... :<

Fable 5: The official doom benchmark is too dangerous to release atm. Will have to wait until society is ready.

Maelstone is a Codex runner for markdown plans. It grinds through one milestone at a time, runs Codex in yolo mode, and refuses to move on unless the work comes back committed, accounted for, and clean. $ maelstone grind --plan plan.md github.com/drdator/maelst…

@sama ChatGPT Image 2.0 is wild. Pasted raw HTML/CSS and asked it to generate a screenshot. Real render vs generated image below. Let's build a browser that renders pages like this.

keyboard_arrow_left Previous keyboard_arrow_right Next

@theo @thdxr Yes the difference is substantial

@shaytan_9 @pshemyslaw I prompt "make a doom-style game using webgl" whenever a new model drops and it's really interesting to see what they produce. 5.5 was quite different compared to other models (fancy 3d, shadows etc).

@simonw My version of this is "make a doom-style game using webgl". GPT 5.5 results are wildly different than previous models

@theo Based on my own highly scientific benchmark it’s about 10% better than 4.6