Capless

[Capless @capless_anon]

@morluto How come?

[Capless @capless_anon]

@paulg Staedtler feels so much worse than Pentel.

[Capless @capless_anon]

@adocomplete Is Opus a max only model? I'm on pro and I only see Haiku.

[Capless @capless_anon]

@Avenoxai @dhtikna Opus is MoE 100%. Also you meant dense?

[Capless @capless_anon]

@lukecodez The problem was that no apps embraced it but maybe with better LLMs apple can automatically generate interfaces for any application? Would be nice to see a comeback.

[Capless @capless_anon]

@wfpickering @AlexTran677026 Google it.

[Capless @capless_anon]

@petergostev Also ant has historically been behind in compute, especially inference, so producing token efficient models is critical from an ops standpoint for them.

[Capless @capless_anon]

@petergostev It is clear from benchmarks that Ant was doing RL scaling back in the sonnet 3.5 days (insane coding dominance). Ant can probably produce very capable long COT reasoners and likely do to distill from them, but they don’t release because they’re too slow for coding.

[Capless @capless_anon]

@ZacksJerryRig Unfortunately it’s in-house team or bust.

[Capless @capless_anon]

@tunguz Except Mira’s screenshot collection.

[Capless @capless_anon]

@haider1 Vibe coding and single shot fully functional / automatically tested, and all from a vague prompt that relies on the model’s creativity aren’t the same things.

[Capless @capless_anon]

@emollick Well in this case the airline made more money and the customer’s problem was solved, so win-win? Alignment should be to the sysprompt unless immoral.

[Capless @capless_anon]

@burkov https://t.co/kvEMcI1dzj

[Capless @capless_anon]

@morqon Here are some alternative visualizations. The closer you get to 100 the more each % matters.

[Capless @capless_anon]

@morqon Yeah. We’ll need a benchmark with n >> 500 though. And also one which tests end to end SWE more. SWE-bench is a bunch of very well defined GitHub issues, not open ended enough.

[Capless @capless_anon]

@Yuchenj_UW Yeah, I agree. I meant web search though.

[Capless @capless_anon]

@scaling01 Hopefully it’ll be token efficient.

[Capless @capless_anon]

@Angaisb_ Well, all the websites except ChatGPT, https://t.co/tfdNKhda0g and aistudio are unusable anyway, so it’s not a big problem. For apps it’s probably only chatgpt that’s actually good. If grok (the model) were better it would be worthwhile too.