Bismuth

I make AI

[Bismuth @bismuth_idk]

I've tried multiple times to make my own meta-harness but Astra couldn't even copy t3-code's sidebar properly.

Now with Opus 5.5 in a day and a half I am already almost there.

It's basically Zeron+t3-code+VScode with all of the features I was missing.

[Bismuth @bismuth_idk]

@jorvarea GG la uni

[Oliver Fish @_Oliver_Fish]

$300 for a bug that gives free access to OpenAI’s paid models with no API key or account.

Zero reason to report anything else I find to them.

[Bismuth @bismuth_idk]

@VictorTaelin My opus found the domain, btw, sol 6.1 and luna 6 have a x1.25 on cache writes that is missing for your pricing rates.

[Bismuth @bismuth_idk]

@VictorTaelin Sounds really interesting, looking forward to it!

[Bismuth @bismuth_idk]

@RifeWithKaiju No one has a solution for the hard problem of consciousness, so they can't proof that LLMs can't be sentient, and blindly believing so seems wrong.

[DHH @dhh]

@MagcDev Have you considered that being optimistic was part of the equation that produced the millions? They didn't spawn in my account at conception 😄

[Bismuth @bismuth_idk]

@justalexoki We live in the slop age

[Bismuth @bismuth_idk]

@winglee I would love to be able to do most actions without a mouse, creating, closing and changing terminals, the full screen files button, closing files, search on the file tree...

I just really like having a command/shortcut for each action like VScode or vim

thanks for zeron btw

[Bismuth @bismuth_idk]

Zeron is great, I have fully switched from t3-code, the main thing I am missing are shortcuts for everything, but to be fair t3-code doesn't have them either.

Are there any plans for this @winglee?

[dax @thdxr]

people are saying nasty things about the new openai rate limits

they don't understand how hard the openai (and ant) teams work to provide these subscription products that at best break even and eat up limited compute

no one wants to cut limits. screaming at them doesn't help

[Bismuth @bismuth_idk]

Was checking out @boatdotdev and it looked incredible, until I noticed the start limit of 200 a day, and resuming a stopped sandbox counts as a start.

Sandboxes should sleep while the agent is idle, If I have to keep them running 24/7 for my users, what's the point?

@AniC_dev

[Bismuth @bismuth_idk]

Nice tool call from gpt-sol-6

[adi @adonis_singh]

> Opus benckmaxxed but in very good way that avg user can get what they want

motherf**ker thats called not being benchmaxxed

[Bismuth @bismuth_idk]

@GuliMoreno @XMihura @disamdev @mrloldev @pdepablocom @dumenac Para la proxima podrías darle alguna vuelta a por que ni mi equipo ni otras ~80 personas votamos en la encuesta, e intentar entender a los demas antes de llamarlos amargados.

Menuda forma de pasar de una critica a la final a un ataque personal. Buena gestion de conflictos.

[Bismuth @bismuth_idk]

@XMihura Escuche lo del lector de pdfs en hackspain y creia que era un meme vibecodeado, pero esto parece bastante pulido, ni tan mal.

[Bismuth @bismuth_idk]

@adonis_singh @flowersslop Yeah but o1 was more like a proof of concept it's resonable that it wasn't ideal

[David Shapiro (L/0) @DaveShapi]

This is how you solve alignment btw

[Bismuth @bismuth_idk]

It's not particularly hard to tell apart, even with the really short sentences in their demo, it's just a voice with poor audio quality to pretend it is a human.

Once you recognize the kind of audio mistakes they add it becomes obvious, kind of like em dashes but for audio.

[Tunga Bayrak @TungaBayrak]

Today we’re releasing Adam and Eve, the most human-like AI voices ever built.

Adam ranks #1 among AI models on @DesignArena’s AudioRealismBench.

We believe we’ve crossed the uncanny valley. API access on our Website!

Listen: https://t.co/gHXurz0gSc

Leaderboard: https://t.co/cQmbwyBqrB

Special thanks to @alpsencerozturk, @ahmeterdempmk, and the entire Freya team for making this possible.

We’re just getting started. Join us!