Super

[Super @_superoriginal_]

@chaotictransfem @aliceisplaying it will be so peak

[Super @_superoriginal_]

@theo @Ga_Vasques Can we please give claude models good computer uses that's on the level of gpt models?

[Super @_superoriginal_]

@Ananth7e Man these bots in the comments are crazy istg

[Super @_superoriginal_]

@argofowl Doesn't seem to be working for me...

[Super @_superoriginal_]

@shivamhwp definitely NOT the "the only harness harness", there are TONS of them, but I find t3 code to be the best one

[Super @_superoriginal_]

@Angaisb_ @argofowl 1. It supports projectless chats now
2. There's a toggle to switch it to the old (codex-inspired) sidebar

[Super @_superoriginal_]

@blueemi99 @LexnLin @itsluneri Oh great another browser. Let me know when it's open source.

[Super @_superoriginal_]

@wholyv Curious on how that works. I'm using Opus 5.5 medium and it uses up my session limit with 1 prompt.

[Super @_superoriginal_]

@TokenGremlin If their models are so good why can't they make good products. Dots feels really underbaked, I still can't access it in the mobile app, the new desktop app feels more and more bloated, and I've completely moved away from it, and their getting cooked by anthropic with opus 5.5

[Super @_superoriginal_]

@zen_browser this shit is so peak i love zen so much! FINALLY we get a better history view

[Super @_superoriginal_]

@Cripple42OB I recommend helium or zen (if youre okay with vertical tabs)

[Super @_superoriginal_]

@noisemakerjon @robinebers Genuinely don't understand how the majority got the wrong answer

[Super @_superoriginal_]

@Raynerdtech Not be a bloated unoptimized ram eating piece of shit

[Super @_superoriginal_]

@JamesZmSun I can't access my dot at all in the mobile app, It's only visible on web and in the desktop app

[Super @_superoriginal_]

@theo @ndycode Thoughts on supporting codex/claude cloud threads? Not everyone has a second device

[Super @_superoriginal_]

@ConnorTalksAI Opus 5.5 is cheaper than astra on their their default reasoning levels (also sure)

[Super @_superoriginal_]

@mg @ChatGPT Dots don't drain usage but the threads that they create do btw!

[Super @_superoriginal_]

@Xxi5olc @synthwavedd that artificial analysis bench shows opus 5.5 being way more expensive than astra max, opus 5.5 on medium (its default effort) outperforms astra on max while being cheaper. (though yes you're right 6.1 sol is WAY cheaper and more efficient than both astra and opus 5.5)

[Super @_superoriginal_]

@Xxi5olc @synthwavedd My point is that the cost of using max reasoning effort should be irrelevant as it isn't good for any sort of real work. It's extremely expensive and slow for often similar or WORSE results to high/xhigh. Max reasoning forces the agents to tokenmaxx and overthink

[Super @_superoriginal_]

@ConnorTalksAI The thing is that opus is better than astra (by quite a bit) while also being cheaper i'm afraid