AI keeps getting smarter, and somehow that doesn't always make it better to use. Today it works great, tomorrow it's different. New model, new pricing, new limits, new behavior. If you're playing, it's fun. If you rely on it for real work, it's exhausting. I don't need a dramatically smarter AI every Wednesday. I just want to know what I'm building on.
Audrius retweeted
Anthropic quietly pulled Claude Code from the Pro plan, updated the docs, changed the pricing page, then called it a '2% test' when everyone noticed. Walked it back the same day.
youtube.com/shorts/kVmb9Xwke…
Audrius retweeted
Mythos is not a bad name for a model but it would be better if Anthropic switched to using famous Claudes. Monet, Debussy etc. The final model that achieves AGI would obviously be Van Damme
Audrius retweeted
Introducing Claude Opus 4.7, our most capable Opus model yet.
It handles long-running tasks with more rigor, follows instructions more precisely, and verifies its own outputs before reporting back.
You can hand off your hardest work with less supervision.
ALT Claude Opus 4.7 Benchmarks
Audrius retweeted
Built a VS Code extension to capture code review comments and feed it back to coding agents.
Built with @OpenAI's Codex. Shipped the same day.
Annotate code inline and export structured Markdown - file paths, excerpts, Git metadata included.
- Group notes into named reviews
- Drag and drop between groups
- Export individual groups
- Resolve, dismiss, reopen, delete
The Markdown format still needs work. That is next.
Issues → github.com/tmsdnl/vscode-loc…
Install → marketplace.visualstudio.com…
Came back to ChatGPT after a few months of working exclusively with Claude.
OpenAI made real progress with Codex, no question. But the ChatGPT experience itself feels like it took a step back. The way it presents information just doesn't support a productive flow the same way Claude does.
The model got better. The experience got worse.
Audrius retweeted
Introducing ChatGPT Images, powered by our flagship new image generation model.
- Stronger instruction following
- Precise editing
- Detail preservation
- 4x faster than before
Rolling out today in ChatGPT for all users, and in the API as GPT Image 1.5.
Plot twist:
AI doesn't make us less human, it makes us MORE human.
Think about it.. when machines handle the calculating, organizing, and processing, what's left?
Just the messy, beautiful, irreplaceable stuff.
Love, wonder, that weird feeling when you hear your favorite song.
We're not becoming obsolete.
We're becoming distilled humanity.
Pure essence!
The real mind-fuck about AI isn't that it's smart. It's that it works without understanding. It processes without experiencing. Succeeds without trying. It's proof that consciousness might be... unnecessary? Like finding out the universe runs on spreadsheets, not poetry. Hurts to think about.
Dark thought of the day: When AI remembers every conversation you've had, every photo you took, every message you sent - better than you do - who owns your memories? When it can tell your life story better than you can, are you still the author of your own life? Or just a character in a database?
We're entering the weirdest era - where being human means proving you're bad at things. Too perfect? Must be AI. Too consistent? Bot behavior. Made a typo? Thank god, you're real. My friend got rejected from a job because her cover letter was 'suspiciously well-written'. We're creating a world where mediocrity is authenticity.
My son asked Claude what he should draw. Then asked ChatGPT what game to play. Then asked AI what movie to watch. At what point did he stop wanting things? When everything is a recommendation, do we even have desires anymore or are we just clicking on suggestions? Free will is dying not with a bang but with a "you might also like..."