Anoop

Never say never

[Anoop @anoop_331]

@reach_vb @remilouf Better model - fast and smart, like opus-5.5

[Anoop @anoop_331]

@bindureddy Why not gpt-6-luna ?

[Anoop @anoop_331]

@reach_vb Can you make this the default in codex cli like Claude does ?

[Anoop @anoop_331]

@scaling01 Benchmarks aside - real world coding it’s way way ahead of GPT models - it’s the combination of smartness and speed. Could be that Anthropic is lavishly throwing limits on subs. Astra on 3D is very good but on coding it doesn’t come close.

[Anoop @anoop_331]

@thsottiaux Opus 5.5

[Anoop @anoop_331]

@thsottiaux @_ak_111 Can you make it ultra good, before making it ultra fast? Opus 5.5 ultra good and ultra fast from the day of release.

[Anoop @anoop_331]

@mark_k Probably another bot slopware. They need to focus on shipping SOTA models like ant

[Anoop @anoop_331]

@VraserX If they have way more capable models internally you would see the effect of that on public models - they are are a projection of what is there internally. If you use Opus 5.5 side by side with current OAI models for real world coding you'd see the stark diff.

[Anoop @anoop_331]

Opus 5.5 >> Sol 6.1 and Astra, OpenAI shipped a bunch of slop, Anthropic with razor focus on model is way ahead.

[Anoop @anoop_331]

@LearnOpenCV After using opus 5.5 it’s really hard to switch back to codex. The api pricings are pretty good though.

[Anoop @anoop_331]

@mark_k I hope not - then it’s kind of very bad deal for api customers, because they marketed sol6 at 50% of price of sol 5.6.

[Anoop @anoop_331]

@jun_song Let’s wait a couple of weeks and see , right now it’s too good to be true, it runs tons of sub agents and the token spend doesn’t move at all. Unleee Anthropic invented some supper efficient inference magic.

[Anoop @anoop_331]

Anthropic is throwing away compute and giving the best of opus 5.5 to get the users they lost to codex during slopus period - make merry of it, while it lasts :-)

[Anoop @anoop_331]

@paolino Move to Sweden ;-)

[Anoop @anoop_331]

@theo Honestly typing - probably because of all the years of coding - somehow makes the thinking that goes with it better. When you talk may be because it’s too fast it’s more instant tokens than the thinking tokens from your brain - it’s just not the same.

[Anoop @anoop_331]

@yoctopokey @GergelyOrosz Who tf is talking about DAU here - we are taking about models capabilities of picking up an existing stack and converting to another, and they will do it better than anything as it has the context. retardmaxxing lfmao

[Anoop @anoop_331]

@yoctopokey @GergelyOrosz I don’t just think, I have done that. Let’s just a wait a few months see how many apps get ported to native.