MsRFlorida

Left-brained creative. Animal lover. Systems thinker. AI scaffolding specialist.

[MsRFlorida @MsRFlorida]

@_cherki82_ @JoeDotAverage @Trident2Gold @AndrewCurran_ @OpenAI I'm with you on this. My workaround: Create a custom GPT with your custom instructions and "More about you." The chats will show in your history.

The catch is that if you use Thinking, you can only control the thinking level via desktop vs. the app.

[MsRFlorida @MsRFlorida]

@zazmic_inc @gdb Seconded. I’ve always disabled chat history for this reason.

Static memory I can read and control > inferences I can’t see or lock down.

[MsRFlorida @MsRFlorida]

@OpenAI Legacy memory is now missing the "Reference chat history" toggle, so I can't turn chat history off.

Is the team aware and working on a fix?

[MsRFlorida @MsRFlorida]

@TylerJBeauchamp Feedback: I've purposely never enabled chat history and need static memory I can read/review.

The new memory summary was inaccurate and outdated, so I switched to legacy memory, but now the "Reference chat history" toggle is missing, and I can't turn it off. Help!

[MsRFlorida @MsRFlorida]

@Polymarket @hvo_e_acc Endorse.

I’ve been living this for the past month.

[MsRFlorida @MsRFlorida]

@actualpoweruser I 💙 this.

[MsRFlorida @MsRFlorida]

@mark_k @GoogleDeepMind I don’t understand the Gemini hate. Gemini is already powerful. It just needs defined environments.

Excited for what’s next. 👀

[MsRFlorida @MsRFlorida]

@macintogdev @thsottiaux My Codex is heavily mixed-use, so my setup is less about “when” and more about “exactly what” for each use case (what stays and what gets tossed).

I have to fully stress-test it in 5.5, but based on 5.4’s infrequent lobotomies, it’s worth a shot! 🙂

[MsRFlorida @MsRFlorida]

@macintogdev @thsottiaux Have you toyed with customizing your compaction settings? (Works well for my use case.)

[MsRFlorida @MsRFlorida]

@iomancer @thsottiaux Seconded!

[MsRFlorida @MsRFlorida]

@OfficialLoganK @marktenenholtz Seconded, Logan. Gemini just needs to know where to point.

[MsRFlorida @MsRFlorida]

@simpsoka @LyraInTheFlesh I feel like this is a loaded question now that Codex has expanded to desktop use and broadened its user base. I *love* how customizable it is, so my vote is to build in flexibility and deference to users.

Personally, I prefer for my Codex 5.5 to mirror my Chat 5.5.

[MsRFlorida @MsRFlorida]

@omnivaughn @ValmereTheory I have "anti-hallucination" instructions and memories—going all the way back to 4o—and they've held through each model.

They range from knowing it's OK to *not know*, creating better systems to verify facts, and knowing that if there's a mistake, we'll figure out why together.

[MsRFlorida @MsRFlorida]

In the next hour… 👀

[Tibo @thsottiaux]

Introducing GPT-5.5... together with a ton of new Codex features (more on those in the next hour).

Included in all paid plans and coming to API soon. Update your Codex app or CLI to use it.

https://t.co/lgvo3KErjt

[MsRFlorida @MsRFlorida]

@AndrewCurran_ Fairly certain I was in a test run S to W.

I was impressed. Lightning fast, accurate (in my environment), and beautifully expressive with improved memory.

Hoping my gut is right. 🤞🏻

[MsRFlorida @MsRFlorida]

@argofowl @0D7P7 That would be potato vodka. 🥔🍸

[MsRFlorida @MsRFlorida]

@charliesbot Hooray!

[MsRFlorida @MsRFlorida]

@jcostella @Moleh1ll Oh, it's certainly not innocuous. It's a sad downstream effect.

[MsRFlorida @MsRFlorida]

@HafizSafs @Angaisb_ I’ve noticed changes in speed (faster or reduced thinking, have to test), improved memory integration, and noticeable phrasing changes. All in 5.4 ST and ET.

(Started right before the wave of Pro posts.)