Miłosz Karolczyk

Hello everyone! Let us know what system architecture is mainly used in your projects. #architecture #monolith #microservices

@claudeebum Do not let us down and keep building!

Got 4000 github nearly in 3days stars after one comment in x what's happening to me

@GohilHardy With Cursor, you effectively pay the API prices for the underlying models, whereas Codex, for example, has a very generous plan. I use Cursor all the time (I have the cheapest plan), but mainly for small things; Codex is my workhorse.

genuine question if Cursor already gives you every Claude model every Codex model every Gemini model plus its own Composer why are people still paying for Claude or Codex separately? what am I missing?

github.com/lidge-jun/open… for anyone who wants to try it as well :)

The Fable/Mythos myth has just collapsed. It turns out it was not about a revolution in models, but a change in the billing model: from API costs to a subscription. Several similar services will appear shortly - and that is a good thing. The winner will be the one that offers the best models, the highest limits, and a predictable subscription. Users do not want to count tokens. They want to pay once and simply use it.

Dear Anthropic, I’m exactly the kind of customer you should love. Sure, I’m only on the Pro plan rather than Max, but I barely use it—I don’t come close to exhausting my limits because I simply don’t need to. Until now, though, I at least knew that whenever I did need you, I’d have access to your best model. And now you want to take even that away from us. @claudeai

@claudeai You still don't understand how not to drive users away. Bye Claude...

Beginning July 20, Claude Fable 5 will be included in all Max and Team Premium plans, at 50% of limits. Pro and Team Standard users will continue to have access to Fable via usage credits, and will receive a one-time $100 credit. Demand for Fable has been challenging to predict, which is why we rolled it out to subscription plans in stages, extending access several times as we secured additional capacity.

The ability for a model to spawn new threads in Codex is a real game changer. The model doesn't just delegate work to other agents. It decides what should be split into a separate thread, supervises their progress, sends follow-up instructions, and even stops them when they're going down the wrong path. For example, watching Sol orchestrate Luna models—assigning work, reviewing results, and steering them when needed—feels surprisingly close to how an engineering lead coordinates a team. The only thing I'm missing is giving subagents within those threads the same level of autonomy. Once that happens, the possibilities will be on another level. #codex

Honestly, all these limits — five-hour, weekly, monthly — look like the invention of someone who had too much free time and too little mercy. Complete nonsense! Let's throw them all out, in one fell swoop, with a bang and confetti! 🎉

@sama Unfortunately, for now the experience in languages other than English is mediocre. Please work on this, because the potential is enormous.

i talk to chatgpt more than i type to it at this point new voice model really crossed a threshold

I love Claude and all its quirks. I just pasted a Fable post about The Matrix and an interpretation of one scene from the film. Of course, I was redirected to Opus. I take it Fable considers The Matrix a threat? ;)

@PhiloGroves But then these "adult" people cry on X that the model deleted something for them :) If they are adults, then they should not complain on X :)

People don't like others making safety decisions for them. Safe defaults, sure, but people should be allowed to accept risk. This goes for both features; I have constant data backups to prevent permanent loss and willing to pay more for larger context processing.

Sol Ultra no longer supports Full Access in Codex. Between this and losing the nearly 50% context window boost, not feeling great.

@thsottiaux Removing the 5-hour limit is fantastic. I can finally focus on my tasks instead of constantly tracking how much time I have left. My work is quite irregular—some days I work more, others less.

We've had no 5h limit in Codex plus and pro for a few days. Do you think it is better or are you finding it difficult to manage the usage included in the weekly limit effectively? If we were to make this different, what should it look like in an ideal world?

@daniel_mac8 Have you tested it?

GPT-5.6 Sol is a great model but in my experience Fable is superior wrt raw capability. GPT-6 is the true rebuttal to Fable. I hear it's coming Soon™.

@trikcode All I had to do was ask Codex to look through Claude’s history...

Got rate limited mid-task yesterday and had to switch from Fable to GPT-5.6 Terra. 6 hours of back-and-forth context entirely gone.Lost 90 minutes just getting a different model up to speed on the project goals. Hitting a usage cap at the worst possible moment sucks. We desperately need a universal way to share context across different models.

@mylifcc Luna is very good, but it needs clear instructions. It does not have the same “feel” as Sol.

Gpt-5.6-Luna max的能力和Sol-high一样高，但费用是1/3，都高于5.5 如果不是特别高的要求，使用luna可以获得3倍的总量

To enable user-input requests by default in Codex, add this to ~/.codex/config.toml: [features] default_mode_request_user_input = true You can also enable it for a single session: codex --enable default_mode_request_user_input Restart Codex Desktop after editing the config.