ahmad ghoniem

Member of Token Burning Staff
Hacking & building with AI Agents

[ahmad ghoniem @_aghoniem]

@pqoqubbw Where's the result
Show us a small vid

[ahmad ghoniem @_aghoniem]

@wholyv I've used 40% of my weekly limit ever since the release 💀

[ahmad ghoniem @_aghoniem]

@lydiahallie @lydiahallie can you help with this?

[ahmad ghoniem @_aghoniem]

it talks so much better
doesn't gaslight/cheat it's way through like Opus 5 did sometimes
the usage is much better
however i was wondering when ant mentioned they boosted 5H limit does that mean they kept the weekly limit the same
i.e instead of getting 12 5h sessions / week you get 10 longer ones for the same weekly limit ?

[ahmad ghoniem @_aghoniem]

@jd_tindall The difference is night and day
Rare anthropic win lately tbh

[JB @JasonBotterill]

Man I am much more productive with Opus 5.5 simply because I can actually fucking understand what it’s saying.

[ahmad ghoniem @_aghoniem]

@SurajHegde6260 @lydiahallie not sure to feel good or bad about it tbh

[ahmad ghoniem @_aghoniem]

@leonardo_lemos @mattpocockuk same

[ahmad ghoniem @_aghoniem]

100% redundant haven't used it for two months

[Thariq @trq212]

we’re thinking of killing plan mode and using the shift+tab hotkey to adjust effort levels

I don’t think the models need plan mode anymore, but if you’re a plan mode diehard would love to get your feedback on why

[ahmad ghoniem @_aghoniem]

Golden

[Dan Robinson @danlovesproofs]

jev is fun to play with but you have to be careful shipping it to prod, over the past few days we accidentally racked up $3.55 in usage

[ahmad ghoniem @_aghoniem]

@sophiadew 👋🏻

[ahmad ghoniem @_aghoniem]

@VictorTaelin I have this theory that Opus 5 was overlooked because they were hooked on fable that Opus 5 turned out unusable without tons of skills to steer it the right the way

[ahmad ghoniem @_aghoniem]

If a message gets flagged, the session switches to an older model and keeps going.

In Claude Code, run /model to switch back, or press Esc twice to edit your last message and retry.

Set "Switch models when a message is flagged" in /config, and report wrong flags with /feedback.

Asking it to show its reasoning in the reply can also get flagged.

Ask "Explain why you chose this approach in three sentences" instead.
6/6🧵

[ahmad ghoniem @_aghoniem]

Anthropic published a guide to getting the most out of Opus 5.5 in Claude and Claude Code, written by
@addyosmani
https://t.co/pBaVHHPxMY
Some prompting habits from earlier models now slow it down.
These are the tips worth changing first🧵

[ahmad ghoniem @_aghoniem]

Add this line to research questions and long tasks:

"Mark anything you couldn't confirm, and say where you looked."

The answer then separates what it verified from what still needs checking, and tells you where it already searched.
5/6🧵

[ahmad ghoniem @_aghoniem]

@dedene med is the default for opus 5.5 while the default was high on opus 5