Fabian Muñoz

[Fabian Muñoz @fmkaizen94]

@shaunralston @sama I think calling it 5.5 is a very smart move, even if it's a different underlying model.

[Fabian Muñoz @fmkaizen94]

@Prathkum @robj3d3 Nah... OpenAI is releasing at the same rate

[Fabian Muñoz @fmkaizen94]

@theo That's why SAAS is dying, including Salesforce. Why a company will paid them if now it can be built internally? Anyway, tailor made solutions always works better than any over engineered SAAS.

[Fabian Muñoz @fmkaizen94]

Completely sandboxed agents are almost useless for multi users systems. Actually, sandboxes (with holes) + prompt instructions are working pretty well with GPT-5.5. On top of that disabling exec or curl to prevent running commands or having web access decrease the agent responses quality.

[Fabian Muñoz @fmkaizen94]

Not windows, is working great on mine. Even weeks ago, when you posted the same thing, it was already working fine. I felt it was very slow the first time and it was because i was using windows, but the repo i was working on was living on the WSL side. I moved the repos to windows and it was solved.

[Fabian Muñoz @fmkaizen94]

@koltregaskes @charliwtquirks Very strange, I'm on windows 11 and i never have the issues you mentioned. I've been tokenmaxing my $200 account for the last weeks and never have a problems with the Codex app.

[Fabian Muñoz @fmkaizen94]

@OlympusBets @MikePFrank @Miles_Brundage This is not the same as steer. Steer is to bypass the messages thread queue and send the message while the model is working. This is different than a side chat.

[Fabian Muñoz @fmkaizen94]

@CyberdyneC @RyanFedasiuk But Dario was begging for this type of regulations. They got what they asked for. Also, there's too little information out there to draw conclusion on what's going on.

[Fabian Muñoz @fmkaizen94]

@ishuagra02 @ai_for_success He is like having butt issues 🤣

[Fabian Muñoz @fmkaizen94]

@benhylak Most of the ones you mentioned are now part of the chat, but not as standalone tools.

[Fabian Muñoz @fmkaizen94]

I was prompting my agent to prioritize github when searching the web for inspiration or solutions, now i include https://t.co/PtNqCLXFHr and is paying off.

[Fabian Muñoz @fmkaizen94]

@haydendevs Bro, are you kidding me? We are a small business and this has been unlocking so much for us, that I can't believe others don't find any value. What's your use case?

[Fabian Muñoz @fmkaizen94]

@jxnlco The most powerful skills in Codex or any other agent, are custom built ones, not some skill you see elsewhere.

[Fabian Muñoz @fmkaizen94]

@damnGruz You should create the keep fable movement. You guys are sick... c'mon bro.

[Fabian Muñoz @fmkaizen94]

Dario Amodei was trying to outsmart every other AI company and end up being outsmarted by Trump and his ways to negotiate.

[Fabian Muñoz @fmkaizen94]

I was reluctant to start using the Codex app because my codex CLI + tmux setup was working so nicely. But I was wrong the Codex app SSH connection and the mobile app work are a game changer.

[Fabian Muñoz @fmkaizen94]

OpenAI it's being so quiet lately.

[Fabian Muñoz @fmkaizen94]

@Angaisb_ You are now like a 4o guy

[Fabian Muñoz @fmkaizen94]

@abacaj Or more if frontier accelerate, like seems happening. I think open source LLMs are now at least 1 year behind frontier.

[Fabian Muñoz @fmkaizen94]

@benvargas @LLMJunky @_Sagiquarius_ Love the idea of a group to make in the cloud local LLM possible. Never heard it before, but I'm interested.