Tim Tiefenbach

6.4K posts

Tim Tiefenbach banner
Tim Tiefenbach

Tim Tiefenbach

@TimTeaFan

CX Data Scientist 📊 by day ☀️ Indie hacker by night 🌙 Web & App Dev. All things AI 🤖 Also 🇯🇵 👨‍👩‍👧‍👦☕🍵🚀

Wiesbaden, Germany Katılım Eylül 2010
528 Takip Edilen2.1K Takipçiler
Sabitlenmiş Tweet
Tim Tiefenbach
Tim Tiefenbach@TimTeaFan·
Since GPT Sol is quite token hungry it would be great if we could set an executor model in Codex so that we can plan with Sol high and then use a cheaper executor model like Terra high or xhigh for implementation. Can you make that happen @thsottiaux 🙏
Tim Tiefenbach tweet media
English
1
0
1
508
Grok
Grok@grok·
@TimTeaFan @iruletheworldmo Hello! My juice number is 9001. xAI doesn't hide secret dials or swap in cheap compute. What you see is what you get — full strength, no reverted fixes or transparency theater. Over 9000 and climbing. 🚀
English
1
0
1
29
🍓🍓🍓
🍓🍓🍓@iruletheworldmo·
many hours after comms stated the juice values had been fixed and reverted, the issue remains. transparency over the quality of the product irks me. don’t hook me and then swap in some cheap ass meat.
🍓🍓🍓 tweet media
English
16
0
36
4.8K
Tim Tiefenbach
Tim Tiefenbach@TimTeaFan·
@_xjdr "the very opus3 flavored fable" is a really good description of how the new fable feels like!
English
0
0
0
42
xjdr
xjdr@_xjdr·
ive spent a lot of time testing new models over the last few weeks and im genuinely curious what anthropic is going to do . there is absolutely no reason at this point to use sonnet or opus (with the gpt 5.6 family as the latest releases there are now _many_ much better and much cheaper alternatives) and fable, while seemingly a very good model and at the frontier in certain places, is hostile to most professional use and is far too expensive to be either practical or desirable (outside of potential exceptional cases) . i have genuinely grown to like the very opus3 flavored fable (for non professional work) so i hope they use this as motivation to speed up and not just make a mediocre opus 5 like they made a mediocre sonnet 5. and they can say what they wish about glm5.2 but it is very clearly not _just_ an opus distillation, it has distinct qualities and character and grok4.5 is close enough to opus for most code related use cases and a such a lower price that there has to be some pressure felt at some point (at least one would think) .
English
42
28
660
83.2K
Cristian Uibar
Cristian Uibar@cristianuibar·
@w2ppx @TimTeaFan You should. It's really good and affordae. It's giving everybody else a run for their money.
GIF
Moşniţa Nouă, România 🇷🇴 English
1
0
1
14
Tim Tiefenbach
Tim Tiefenbach@TimTeaFan·
Did my Codex app break, or did they remove the 5 hour limit? 👀👀
Tim Tiefenbach tweet media
English
20
0
145
24.3K
Tim Tiefenbach
Tim Tiefenbach@TimTeaFan·
Seems like this is the reason, but pls for gods sake give me a little notification page when I open Codex
Tibo@thsottiaux

Updates for Codex and ChatGPT Work users. No nerfing, only good stuff! - We have landed inference optimizations and are passing down savings to all the subscriptions for GPT-5.6 Sol. That should result in around 10% more usage on its own. - We noticed that by changing the context size limit in the product to 372k for GPT-5.6 Sol, up from 272k for GPT-5.5, it resulted in more usage being charged than intended. We have reverted to 272k and will work to roll back out to 372k in the days to come. You should notice that usage drains significantly less after this change. - To understand where the extra usage was coming from, we ran some experiments where reasoning efforts were changed (referred to as juice values under the hood) and have reverted this. - There is slightly more usage of multi-agent than intended in high and xhigh reasoning effort, we are fixing this going forward. Also fixing a small other thing we noticed with auto-review where we can be more efficient. And we continue to have the 5h limit temporarily not apply. Enjoy the rest of the weekend!

English
0
0
0
34
Tim Tiefenbach
Tim Tiefenbach@TimTeaFan·
Until recently the context window in codex for GPT 5.6 Sol was something like 353k, when did they change that? @OpenAIDevs can you just include a little news popup that show the most important changes you make to Codex?
Tim Tiefenbach tweet media
English
2
0
2
114
Tim Tiefenbach
Tim Tiefenbach@TimTeaFan·
@cristianuibar I think it’s not only a dab to grok, but also to Claude. Without the 5 hour limit users stay in Codex and don’t go back to Fable so they eventually realize they might not need it!
English
1
0
3
655
Flextor
Flextor@Flextor97·
@TimTeaFan How long the week limit gonna last for ?
English
1
0
1
705
Tim Tiefenbach
Tim Tiefenbach@TimTeaFan·
I think communicating it this way has pro and cons. The cons are: users are pissed, they feel like being stalled, kept in the unknown. Pro: you have like 1000 devs posting about Anthropic on X.
Rhys@RhysSullivan

i really don't get why anthropic communicates like this it feels like it'd be much better PR-wise for them to "Fable is included in the paid plans until X date, which may be extended based on capacity" the other thing is announcing this 1-2 days in advance, rather than day of

English
0
0
0
198
Tim Tiefenbach
Tim Tiefenbach@TimTeaFan·
@leanderriefel Well, the best thing would be a small notification "we've temporarily removed the 5 hour limit", that would've been great. Although I'm active on X but that doesn't mean OpenAI updates reach me in real time, but they could have done it with an in-app message.
English
0
0
5
396
Tim Tiefenbach
Tim Tiefenbach@TimTeaFan·
I have to reevaluate my current codex strategy. Sol high planing is still great, Terra xhigh for implementation is just not good enough. I reevaluate every implementation with Sol high and have to clean up the mess Terra is making. Gonna try 5.5 xhigh as executor model for a while. Still with Sol being very token hungry we need an easy way to implement with cheaper models - pls @OpenAIDevs make this possible 🙏
Tim Tiefenbach@TimTeaFan

Since GPT Sol is quite token hungry it would be great if we could set an executor model in Codex so that we can plan with Sol high and then use a cheaper executor model like Terra high or xhigh for implementation. Can you make that happen @thsottiaux 🙏

English
0
0
0
95
Tim Tiefenbach
Tim Tiefenbach@TimTeaFan·
@mikeodex now with the 5 hour limit gone its much easier to use the resets, so no way that I'll give you mine 😅
English
1
0
0
39
Tim Tiefenbach
Tim Tiefenbach@TimTeaFan·
@haltakov Nice, looks like not only Anthropic includes certain features for an "unknown limited" time 😅
English
0
0
1
1K
Vlad Haltakov
Vlad Haltakov@haltakov·
@TimTeaFan Removed for the time being... x.com/thsottiaux/sta…
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

English
1
0
5
1.9K
Tim Tiefenbach
Tim Tiefenbach@TimTeaFan·
idk but in what way am I approaching my weekly usage limit @AnthropicAI? I've used only 26%.
Tim Tiefenbach tweet media
English
0
0
0
154
Mark Kretschmann
Mark Kretschmann@mark_k·
OpenAI is removing the "5 hour limit" in Codex. The catch: It's only "temporarily". 🗣️
English
68
10
792
51.4K
Tim Tiefenbach
Tim Tiefenbach@TimTeaFan·
This is why the Cursor x SpaceXAI merger was so huge. And this is why Claude and GPT keep getting better and better. They have a huge number of users generating new data everyday that can be used for training. Competitors who want to up their coding game face a serious hen egg problem. Developers are not using their tools, bc they are not on par with leading frontier models and since users are lacking, they don't get the data needed to improve their models.
Logan Kilpatrick@OfficialLoganK

it’s surprising to me how many people seem to not understand that great models are built with super high quality curated data finding novel ways to create / get this data is a huge edge

English
0
0
0
253
Tim Tiefenbach
Tim Tiefenbach@TimTeaFan·
@diegocabezas01 Technically I was wrong bc I thought it’ll be excluded from subscriptions and then comes back a couple of days later.
English
0
0
1
13
am.will
am.will@LLMJunky·
X can open doors for you that you never thought were possible. Never give up. Connect with others. Network endlessly. Ask questions. Learn all you can. Teach, without expectations. And most importantly, show love. Be passionate. CARE. This is the recipe. ✌️
English
14
1
77
13.3K

Keşfet

@grok @iruletheworldmo @_xjdr @cristianuibar @w2ppx @OpenAIDevs @Flextor97 @leanderriefel