This take is just simply dumb. Maybe ai doesnt write the most efficient or fully working code on all edge cases in the first shot. But for sure there was bugs in the past, a lot of them. Currently the only reason why there are more bugs is because there are more projects also.
This is fake.
Before AI there were no bugs.
Before AI there were no security issues.
Before AI there was no slop.
100% of code was high quality and slop was first introduced in 2023 with ChatGPT, but some of you are too young to remember.
We’ve received notice that the Department of Commerce has lifted export controls on Claude Fable 5 and Mythos 5.
We'll begin restoring access tomorrow, and will share an update soon.
We’re grateful to our users for their patience, and to everyone who worked with us on redeploying the models.
@juust@pmitu 1. If you waste the models capability on small tasks, then you deserve to pay the premium.
2. Its more expensive, yes, but you are forgetting the intelligence gap in between the two models
Claude Sonnet 5 just dropped.
Pricing: $3 input / $15 output per million.
Same as Sonnet 4.6.
But the benchmarks tell a different story.
Sonnet 5 now sits within striking distance of Opus 4.8 on every category.
And it actually beats Opus 4.8 on GDPval-AA v2 knowledge work.
You are now getting Opus 4.8 level intelligence at one third the price.
Sonnet 4.6 is officially dead.
Sonnet 5 is the new daily driver.
Honest question:
Is anyone actually making money from these weekend AI projects, or are we all just building apps that only run on localhost and posting screenshots on X?
Exactly. Has to be regulated. Models have already been exploited, and we are in luck that they are “not that intelligent” but imagine fable 5 open sourced. That would be fearful in the wrong hands. Regulate that shit rn
🚨ANTHROPIC CEO: OPEN SOURCE AI IS GETTING DANGEROUS
Anthropic CEO Dario Amodei told lawmakers that open-source AI is moving down a “very dangerous path.”
His warns that once powerful models are released openly, companies lose the ability to monitor misuse, revoke access, or update safety guardrails.
GPT‑5.6 Sol sets a new state of the art on Terminal‑Bench 2.1, which tests complex command-line workflows requiring planning, iteration, and tool coordination.