Jonathan Haas

agents for work @evalopsdev

[Jonathan Haas @JonathanHaas]

@evanyou Huge congrats :)

Excited to see the benefits this has (and hopefully means you get to take some well deserved vacation!)

[Jonathan Haas @JonathanHaas]

@EniacVC why not both?

[Jonathan Haas @JonathanHaas]

@thomasgauvin @Cloudflare where does one cop tho

[Jonathan Haas @JonathanHaas]

@blennon_ Awful messaging

[Jonathan Haas @JonathanHaas]

@Suhail LFG!

[Jonathan Haas @JonathanHaas]

@matanSF @HarryStebbings Cannot wait to listen to this one :)

[Jonathan Haas @JonathanHaas]

@jeffwsurf 🫡

[Jonathan Haas @JonathanHaas]

@surim0n good! super app for all.

[Jonathan Haas @JonathanHaas]

I really wish more people understood this. The hidden assumption in every "use cheaper models to save money" pitch is that task completion is held constant. It's not. Cheaper models generate more tokens correcting their own mistakes, longer tool loops, more retries. The per-token price is lower; the per-outcome price often isn't. I know you know this. I know you say this openly. I just wish more folks could get past the weird mental block that seems to exist :\

[Jonathan Haas @JonathanHaas]

Endless missing model spam on Codex? @OpenAIDevs

[Jonathan Haas @JonathanHaas]

@signulll Uber doesn't need more code. they needed to make sure Google didn't have those engineers. now they don't.

[Jonathan Haas @JonathanHaas]

@thsottiaux @MadisonMills22 you tell me? y'all hiring :)?

[Jonathan Haas @JonathanHaas]

@Dimillian Hell yeah! What a good ship week.

[Jonathan Haas @JonathanHaas]

@natolambert @allen_ai Incredible run. Excited to see whatever is next Huge congrats, Nathan!

[Jonathan Haas @JonathanHaas]

@atShruti Has been super painful for us to use recently :\ Not sure if it's the most recent models or strong degradation of the harness

[Jonathan Haas @JonathanHaas]

@sqs This is a good call!

[Jonathan Haas @JonathanHaas]

@DevenNavani @cognition @DevinAI student confidently raising their hand to explain the homework to the teacher who wrote the textbook

[Jonathan Haas @JonathanHaas]

@_arohan_ there is no escape. there is only required, optional, and repeated

[Jonathan Haas @JonathanHaas]

funniest possible outcome is the AI reads the S-1 announcement, summarizes it neutrally, and adds 'this is after my knowledge cutoff so I can't verify it'" - which, for the record, is exactly what Claude did twenty minutes ago

[Anthropic @AnthropicAI]

Anthropic has confidentially submitted a draft S-1 registration statement to the Securities and Exchange Commission.

Pending completion of SEC review, this gives us the option to pursue an initial public offering.

Read more: https://t.co/onGZAhRLvD

[Jonathan Haas @JonathanHaas]

@ivanburazin Yeah, that's GPUs George. We feed him red bull.