mark erdmann

5.3K posts

mark erdmann banner
mark erdmann

mark erdmann

@markerdmann

building https://t.co/FSCuz2oofs - the future is air conditioned bytecrafter humbly serving atomcrafters prev cofounder @pulley

Calgary Katılım Aralık 2008
2.1K Takip Edilen1.4K Takipçiler
mark erdmann
mark erdmann@markerdmann·
or two apps! it's so annoying to tap your way out to ask a quick chatgpt question, then tap your back in to check on your codex threads, then tap your way back out to see the answer to the quick question. basically have to choose either chatgpt or claude as the quick question app, depending on what you're currently using for coding agents.
English
0
0
2
25
mark erdmann
mark erdmann@markerdmann·
@randal_olson Kinda feels like they have a secret meta-effort dial that they can adjust according to capacity
English
1
0
23
5.4K
Randy Olson
Randy Olson@randal_olson·
You can tell Claude Opus 5 is coming soon because Opus 4.8 is SO BAD today. I'd love to know why the current models degrade so badly the day before they do a new release.
English
63
8
537
64K
mark erdmann
mark erdmann@markerdmann·
@skeptrune ah same here, so i use a voice agent and mostly look out the window. works beautifully.
English
0
0
1
26
Nick Khami
Nick Khami@skeptrune·
waymos are very slow to get places i'm directionally bearish on founders or otherwise supposedly busy people that swear by them feels unserious
English
59
0
132
20.7K
mark erdmann
mark erdmann@markerdmann·
told a non-developer friend about ChatGPT Work and pasted the link - not confusing at all 😂
mark erdmann tweet media
English
1
0
1
135
mark erdmann retweetledi
Adrien Ecoffet
Adrien Ecoffet@AdrienLE·
Kinda cool that tibo can reset the usage limits at ant now too
English
19
53
2K
86.7K
mark erdmann retweetledi
0xSero
0xSero@0xSero·
GPT-5.6-Sol (medium) is the optimal configuration.
0xSero tweet media
English
58
23
677
62.1K
Anthony Kroeger
Anthony Kroeger@kr0der·
it's been 1 day now - what reasoning level is everyone using for GPT 5.6 Sol?
English
278
4
296
72.4K
Tulsi Prasad ⚡️
Tulsi Prasad ⚡️@thebuildguy·
@markerdmann very interesting experiment! i'm curious what fable has to say about all of gpt models now! 😅 let us know if you see difference between sol (w differnt effort lvls) vs sol,tera, luna
English
1
0
1
29
mark erdmann
mark erdmann@markerdmann·
more like gpt-5.6-sold-out amirite
mark erdmann tweet media
GIF
English
0
0
0
83
mark erdmann
mark erdmann@markerdmann·
codex has this annoying new bug where it stops working and needs a nudge or maybe this is a bug in the custom subagent orchestration that sol added for me yesterday? cc @thsottiaux
mark erdmann tweet media
English
0
0
0
68
mark erdmann retweetledi
Matthew Berman
Matthew Berman@MatthewBerman·
DeepSWE says GPT-5.6 Sol High is the best overall (price/performance) model on the planet. I agree.
Matthew Berman tweet media
English
40
26
603
29.9K
mark erdmann
mark erdmann@markerdmann·
@seconds_0 yes, hallelujah. sol is light years ahead of 5.5 in this regard. and dare i say better than fable? it seems less repetitive in its writing?
English
0
0
6
568
0.005 Seconds (3/694)
0.005 Seconds (3/694)@seconds_0·
sorry the greatest benefit to gpt 5.6 sol is that it is frontier intelligence that just talks normal
English
13
9
335
11.3K