Francesco Vaccaro

[Francesco Vaccaro @Smart9777]

https://t.co/YoS12rk9J1

[Francesco Vaccaro @Smart9777]

GPT 5.5? While we have 5.6 Sol and Kimi K3?

This move reminds me of this

[GitHub @github]

📣 GPT-5.5 this weekend: 60% off for Copilot Max users and 40% off for Pro+ users. 🎉

Put it toward a tricky bug, a larger change, or updates across multiple files — wherever you use Copilot.

No code or opt-in required. Happy coding. ✅

[Francesco Vaccaro @Smart9777]

@thsottiaux Much much better, i can better manage my weekly limit. Sometime I can use codex the last days before the weekly reset and because of the 5h limit i can’t ever finish my quota

[Francesco Vaccaro @Smart9777]

@scaling01 But mythos was also "updated" in april while it was in project glasswing. For sure OpenAI and Anthropic are internally in the lead but idk if atm they are leading 7-8 months on all areas.

I think that K3 could be about 2-3 months behind in code and 7-8 in other areas.

[Francesco Vaccaro @Smart9777]

I love how GPT 5.6 Sol is faster than 5.5, it produce less output token and howewer it’s very reactive at steering.

Bonus, this is what i built with it 😁

[Francesco Vaccaro @Smart9777]

https://t.co/YoS12rk9J1

[Francesco Vaccaro @Smart9777]

No one talk about that but I love the way GPT 5.6 Sol is much more responsive to task steering than GPT 5.5.

Instead Fable 5 it’s very difficult to steer coz it never shows you the reasoning steps while working on a task, I hate this behaviour, I hope they can fix it with 5.1.

[Francesco Vaccaro @Smart9777]

@OpenAI pls copy this feature, if you can with personal and enterprise/work account. Thanks

[Computer @AskPerplexity]

The Perplexity Mac app now has an account switcher.

Switch between your saved accounts, add a new one, or manage them from the sidebar profile menu or the account settings menu.

Live now in the Mac app for all Computer users.

[Francesco Vaccaro @Smart9777]

@Carboaceto Maybe they’ll have a different set of problems to solve, problems we can’t even imagine right now because we’re in a transitional period.

At the moment, when we train someone new, we start with raw code and then move on to spec-driven development.

[Francesco Vaccaro @Smart9777]

@mweinbach Oh nice. I have to re run my tests coz yesterday it was using less token than 5.5 but eating about 17% more of my 5h window quota, at same reasoning level (medium) on my plus plan

[Francesco Vaccaro @Smart9777]

@thsottiaux I was one of the user affected. Thank you so much, you and OpenAI one of the best big companies at hear community feedbacks

[Francesco Vaccaro @Smart9777]

@kimmonismus I tested the efficiency with codex telemetry and 5.6 Sol medium produce 34% less token than 5.5 medium but the eats the usage quota of the 5h window of my plus account about 17% more.

There is something strange

[Francesco Vaccaro @Smart9777]

So… maybe something was really off with usage limit of 5.6 Sol

[Tibo @thsottiaux]

Morning. The last 48 hours of Codex and ChatGPT Work have been intense! Three important updates:

- Temporarily removing the 5 hour usage limit restriction for all Plus, Business and Pro plans
- Rolling out changes that will make GPT 5.6 Sol more efficient across the board and that will be reflected in less usage being used so that it can take you further. Exact impact to be quantified and shared
- We hit 6M active users, and are landing a usage reset in the next hour

Go do things

[Francesco Vaccaro @Smart9777]

@sickdotdev I'm made a benchmark to try to verify this "sensation". I found that 5.6 Sol while using less token, eats more quota of the 5h window.

Website: https://t.co/Mx9UtM2Oi6

Github repo: https://t.co/9L3efbRpry

Article: https://t.co/wGUJaUup3l

[Francesco Vaccaro @Smart9777]

https://t.co/YoS12rk9J1

[Francesco Vaccaro @Smart9777]

@thsottiaux here you (and anyone else who can try it) can find website e Github repository of the benchmark cited in the article.

Website: https://t.co/FKsIWvEhPz

Github repo: https://t.co/9L3efbRpry

I hope you find this useful

[Francesco Vaccaro @Smart9777]

https://t.co/YoS12rk9J1

[antirez @antirez]

It is my belief that many devs right now are not maximizing what they can do with automatic programming because they still look at the code. Doing it makes you the bottleneck. Your time is better invested in new ideas, QA, design, and asking yourself what is your goal.

[Francesco Vaccaro @Smart9777]

@cheatyyyy @thsottiaux @reach_vb Same issue here🥲

[Francesco Vaccaro @Smart9777]

@kimmonismus I can confirm. Something is off while using 5.6 Sol. I tried to verify this issue and I found that with less output token 5.6 Sol medium eat more usage of my 5h window, than 5.5 medium.

I tried to document my tests here https://t.co/VLyNxkFPp1

[Francesco Vaccaro @Smart9777]

https://t.co/YoS12rk9J1

[Francesco Vaccaro @Smart9777]

@thsottiaux I may have found evidence that GPT-5.6 Sol actually burns through Codex quota faster than 5.5.

I used 5.6 Sol itself to build a benchmark, and these are the findings, assuming the methodology has no major flaws 😅

Happy to share thread IDs, raw data, or any other details that could help. Also repo is coming soon.

[Francesco Vaccaro @Smart9777]

https://t.co/YoS12rk9J1

[Francesco Vaccaro @Smart9777]

@_lia_giu_ Ok boomer. 🎩