AI model launch frenzy (some better than others) is making a tool like T3 Code start to make sense to me.
After being Cursor-only for a long time, I'm running Claude Code side by side now. Opus 5.5 feels a magnitude ahead, especially next to Grok 4.7.
I picked Cursor because it could run Anthropic, OpenAI, and other models in one place. I hate wasting time configuring and syncing multiple tools for the same job. Besides that, each will still have its own specifics.
Finally got tired of macOS naming everything https://t.co/k5l5nF3md1.
Had Grok Bot write me a script that names the ZIP after the first file I select, and skips the __MACOSX junk. Two steps gone. I can invoice clients like a human again.
Cursor shows this on day 2. But after putting it through my daily workload and beyond, I feel like going back to 4.6 and leaving 4.7 for the PCB side-project.
This was actually good. And code review flagged some things 4.6 did not catch.
BUT... Opus 5.5 caught them even more. Putting it through today's pesky JFR files as I type this.
@DanielLockyer Remember the days when software had to work (and it did indeed) on far lower hardware specs than nowadays' fitness bands? What happened?
You ever notice how each model builds its own canvas style for the outputs? It would be great if a predictable canvas render handled different models' output the same way.
If I get my first 3D printer (something like a Bambu Lab P1S), what do I actually need to know about ultrafine particles and VOCs?
I see they advertise PLA Pure filaments, but how much safer are they actually? Asking so I can decide whether to bring out the big rig and drill an exhaust hole into the wall.