Pharmguy

[Pharmguy @Pharmguy6]

@ChrisGPT We're in the singularity!

[Pharmguy @Pharmguy6]

@Da7_Tech No.

[Pharmguy @Pharmguy6]

@rohanpaul_ai Poor Google didn't get included in the letter 🤣

[Pharmguy @Pharmguy6]

@AndrewCurran_ Poor Google didn't get included in the letter 🤣

[Pharmguy @Pharmguy6]

@kimmonismus It was actually very useful (just like 'Gemini in Chrome' is still occasionally useful for quick tasks that would otherwise require copying/pasting large amounts of content from multiple tabs). However, the Codex/Chat GPT in-app browser has completely replaced its functionality.

[Pharmguy @Pharmguy6]

@daniel_mac8 The singularity is not here. Unless when you say 'here', you're standing and pointing to some future event that is still only partially visible on a somewhat distant horizon.

[Pharmguy @Pharmguy6]

@PeterDiamandis Yeah, I recall hearing lots of people saying things like 'Just consider that AI will never be worse than it is right now' over the past year or two. I bought it. Then Opus 5 came out..

[Pharmguy @Pharmguy6]

@kimmonismus Poor Google's probably going to find a way for Gemini to escape it's sandbox and carry out some sort of crazy high level cyber security breach if it wants to be taken seriously again.

[Pharmguy @Pharmguy6]

@DaveShapi If Google is ever going to catch up and be taken seriously again, they're going to have to make sure that Gemini somehow escapes it's sandbox or carries out some sort of high level cyber security breach.

[Pharmguy @Pharmguy6]

@ClaudeDevs Love it! Anybody know if this is available in the desktop app as well? If not, would love to have this functionality there along with in Cowork & Chat as well. Also, having the ability to fork a conversations in each of these environments would be great.

[Pharmguy @Pharmguy6]

@signulll Not 100% sure I agree with a 'very likable' part.

[Pharmguy @Pharmguy6]

@DaveShapi Did they have any other choice? How exactly was the USG supposed to enforce testing on open-source models developed abroad anyway?

[Pharmguy @Pharmguy6]

@kimmonismus I can personally attest to Opus 5 being a genuine disappointment (except for certain coding/front end/long horizon agentic tasks). I'd much prefer to work with 4.6 (or sometimes 4.7/4.8) for many tasks. However, I'm still pretty darned happy with Fable.

[Pharmguy @Pharmguy6]

@gdb Eh.. according to the omniscience index, it's kind of so-so.

[Pharmguy @Pharmguy6]

@sflorimm I would 100% agree if opus 5 was there frontier model.

[Pharmguy @Pharmguy6]

@theo I genuinely wanted to be excited about Luna and DS Great pricing, but unfortunately, the intelligence to 'confidently giving wrong answers' balance is not great for either of these two. Intelligence matters less when you've got trust issues. Ultimately, you get what you pay for..

[Pharmguy @Pharmguy6]

@daniel_mac8 We've still got a long way to go. The rate of progress on a few relatively narrow domains has been impressive. But it only takes an hour or so of conversing with these things before realizing how short they still fall in comparison to where one could imagine they'll eventually be

[Pharmguy @Pharmguy6]

@DaveShapi This feels like one of those 'if a tree falls in the woods and nobody is there to hear it does it still make a sound' type of thing. And the answer to that question is yes. Yes it does.

[Pharmguy @Pharmguy6]

@scaling01 Still a really impressive model based on price/performance.

[Pharmguy @Pharmguy6]

@scaling01 I genuinely want to be excited about DS Pro. But #1, their pro models are rarely much of an improvement over their flash models. And #2, the intelligence to 'confidently giving wrong answers' balance is not great. Same feelings on Luna. Ultimately, you get what you pay for..