Potato

@ns123abc This was test to identify the token usage. Tibo already confirmed to have reverted. Stop spreading misinformation

🚨BREAKING: OpenAI has nerfed GPT 5.6 SoL >just 4 days after release It’s over

@lfji @nickreese @thsottiaux @mylifcc Are you using Codex in the cloud? Codex in cloud burns tokens really fast.

@SaladHug @nickreese @thsottiaux @mylifcc > I never succeeded using all my tokens for a week and I'm a daily codex user That’s odd. I can burn through 97% of my weekly limit in just 24 hours with gpt-5.6-sol medium (not `fast`).

找到一个降低5.6Sol token消费的办法： 在~/.codex/config.toml 中加入 model_context_window = 272000 model_auto_compact_token_limit = 240000 原理下面说：

@AArdvarkErick @thsottiaux All complaints are customer feedback!

@thsottiaux Calling customer feedback “complaints”. Bruh.

At different times you complained about speed, code slop, frontend quality, ... With each release we improve and GPT 5.6 Sol is ✅Fast and token efficient ✅Hardcore at back-end dev ✅Great at front-end ✅Does not use useEffect everywhere What is next?

@nickreese @thsottiaux @mylifcc Are you using ultra? I think it only drains fast on ultra, especially when using Fast. Otherwise I don't really see a big difference.. what are you doing to drain it so fast? I never succeeded using all my tokens for a week and I'm a daily codex user

@thsottiaux @mylifcc 5.6 drains suuuper fast... way more than 5.5 something is up.

@kimmonismus Are you using ultra with super speed? It's the only way I get the 5 hour limit. For now I hit it 3 times but 2 times right before the 5h reset. Without ultra this doesn't happen and I use medium/high

Another reset, which is unfortunately necessary as well. Unfortunately, it is becoming increasingly clear that the token burn rate is simply insane. I have never come close to hitting the 5-hour and weekly limits this quickly on the Pro tier as I have with 5.6 Sol. I mostly use High, not Fast Mode. 5.6 is undoubtedly better. The improvement is most noticeable in frontend work. That was where I immediately noticed that they had genuinely caught up. 5.6 also works better and remains agentic for longer. I am very satisfied overall. It is a significant step forward, if it were not for the rapid consumption. I also still do not understand the “super app” concept. Why should I use the “Work” environment when Codex does exactly the same thing while also displaying more information? And why do I still have the classic ChatGPT app? Perhaps I missed something, but at the moment this is once again creating more confusion. I am still hoping for resets, because otherwise this will become difficult. As for the claim that 5.6 is 54% more token-efficient, I consider that made up. It does not align with my usage at all. In fact, the opposite seems true. It feels more like 54% less token-efficient. So, 5.6 is a good update with one major caveat.

@danielchu83 @thsottiaux I think it's fine. What you want may translate to have Sol ultra be replaced by Sol medium for example and rebranded as ultra. More usage for sure but less intelligence for everyone because too expensive. I don't mind having access to expensive models, that's what I pay for.

@thsottiaux I think the “definition” and hence the design and implementation of Ultra may not be correct. Even at Pro sub, one can probably only have 2-3 Ultra sessions. Ultra shouldn’t have been designed as “maximize token burn regardless of outcome”.

Hello beautiful people! We have reset usage limits across Codex and ChatGPT Work. And another one will come later in the day. Rejoice. Now that I have your attention, a quick update on ChatGPT Work, Codex and all the updates we shared yesterday. We’ve spent the last 24 hours reading feedback, looking at usage patterns, and talking with many of you. The short version is that there is a *lot* of excitement for GPT 5.6 Sol, ChatGPT Work on mobile & web, but also that we didn't get everything quite right. - We made it too easy to use the highest-compute settings without making the impact on usage limits sufficiently clear. - We reorganized the desktop app in one bold move, making familiar things like chats and projects harder to find. - Our launch framing was focused on ChatGPT Work and to some of our Codex fans it made it feel like Codex was going away over time. Absolutely not our intention, we love Codex and it is here to stay. - And we introduced regressions for some existing multi-agent workflows, alongside a collection of rough edges in plugins and other parts of the experience. We’re landing a first set of improvements today. We’re resetting usage twice so people can keep experimenting, changing defaults and the model picker so they don’t push people toward unnecessarily expensive settings, fixing several plugin submission issues, improving how we represent Codex in the product, and cleaning up some of the most immediate desktop problems. A larger set of improvements will land next week. We’re bringing chats and projects back into the sidebar in a more familiar and customizable way, making usage and reset timing much more visible, clarifying when to use ChatGPT Work and when to use Codex, and addressing the many other smaller pieces of great feedback we've had. The ambition behind this launch hasn’t changed. We think bringing ChatGPT and Codex together into a workspace where people and agents can collaborate is a very important step forward. But an ambitious direction doesn’t excuse avoidable confusion or regressions in the first version. Please keep the feedback coming. We’re moving quickly, and you should see the experience already get better with a few updates today; and substantially better again next week.

@GithubProjects I'd like it if it didn't outright fuck the contrast on the MacBook Air. Native mpv with a nice osc (and there are lots of those) works better for me

IINA is a modern macOS video player built on mpv, designed for performance and native integration with recent versions of macOS. - Force Touch, picture-in-picture, and advanced Touch Bar support - Online subtitle searching and intelligent local subtitle matching - Standalone Music Mode for audio files with video thumbnails - Fully customizable keyboard, mouse, trackpad, and gesture controls Explore it here: osp.fyi/iina

@rohntwts @ChatGPTapp Unfortunately for same reason Excel only has a similar feature for US only. EU regulations.. Was really looking forward to this in Excel years ago, still ain't there. I want this so bad in chatgpt, but I won't expect it to ever come.

@ChatGPTapp Why only to the users in the US? Do you think only people in US have money?

Questions about dollars. Answers that just make sense. Personal finance in ChatGPT is now available to Plus users in the U.S.

@thsottiaux @iamsahaj_xyz I've been doing this myself this way. I'm curious, do you use an instruction in agents.md for this?

@iamsahaj_xyz I personally DM it to Codex and it stores it into my notes folder all neatly organized. Some things it then just does itself and for the rest it reminds me of things I need to do a few when I have free time based on my calendar

dm'ing yourself on slack is the best way to take notes. fight me

@PCMR_Elitist Once Nvidia is supported I'm doing a 180 from Windows to SteamOS

So, who's all planning to switch to SteamOS and leave Windows behind?

@FlowAltDelete The meeting transcriptions by itself is surprisingly bad.. maybe it's because in our company we limit computing to EU. But there ain't a world where I can confidently say I'd choose Copilot over Chatgpt, Claude, Grok or Gemini. Yes I've tried them all extensively

@SaladHug Not true.. Just by having Copilot inside Teams meetings is worth 30$. At least for me it is.. Maybe I have too many meetings

People are acting like Microsoft robbed them. NO. Microsoft let us hammer Copilot Cowork for free since March. - Long-running tasks. - Multi-step work. - Emails. Docs. Meetings. Files. Context. Execution. Some of us were absolutely hammering it too. I was one of them. Now GA hits and suddenly everyone is shocked there’s a meter attached? Come on. The move now is simple: >stop wasting context >stop writing messy prompts >stop asking Cowork to do 9 vague jobs at once >start building proper SKILLS >start optimizing how work gets handed off Microsoft gave us the playground. Now it’s time to operate like adults. I'll be posting tips and tricks to keep costs down. Keep an eye out 👀

@jxnlco I wish we could see credentials linked to our plugins, even better would be the ability to have more than one. Other than that I think they're useful, I even created one for an ITSM tool we use at work.

codex users! two things i want feedback on: are plugins actually making codex better? which ones, and what still feels broken? how are you using codex as a team? drop examples and i’ll organize them for the team!

@kimmonismus Yeah, I have the HM Aeron and the HM Embody. Expensive af, but I sit a lot and I see it as an investment for the good health of my back. Both are great, I prefer the Aeron but you can't cross legs. Embody is hit or miss, hit for me. In summer the Aeron is better because of sweat

I need a new office chair. Ive heared Herman Miller is the goat. is that true or do you have other recommendations?

@NVIDIAGeForce #RTXPowersPlay

To celebrate #SummerGameFest... We have a GeForce RTX 5090 up for grabs👀 Want it? Comment #RTXPowersPlay to enter.

@ajambrosino Now you're saying something @ajambrosino.. thanks for the tip. I was thinking about a bigger scale of this for a moment, but if everyone has codex and they work.. they will end up fixing things by themselves anyway

install codex on your parents’ computers so you can fix stuff remotely

@NVIDIAGeForce #RTXOn let's go!

Over 1,000 RTX games and apps are available now with ray tracing and DLSS. To celebrate, we're dropping Steam cash in the replies to upgrade your library with a #RTXOn title... Comment #RTXOn to enter 💸

@Govindtwtt But google gives me an AI answer

STOP ASKING CHATGPT THINGS YOU CAN GOOGLE. STOP ASKING CHATGPT THINGS YOU CAN GOOGLE. STOP ASKING CHATGPT THINGS YOU CAN GOOGLE. STOP ASKING CHATGPT THINGS YOU CAN GOOGLE. STOP ASKING CHATGPT THINGS YOU CAN GOOGLE. STOP ASKING CHATGPT THINGS YOU CAN GOOGLE.

@DanielatOCN Thanks, I needed this years ago and still do today. I don't need it often, but I appreciate this so much!

I've collated 2400+ Microsoft architecture icons and published them for easy access at msicons.com 💙 For every icon, you can: • Download the scalable SVG file • Download the PNG of your desired size • Embed the icon directly into your website or markdown • Choose a light, dark or transparent background You can can even compare different icons together and favourite multiple icons for bulk download! #Microsoft

@fandompulse Fuck wokeness and all that honestly, but James Bond always has the boss lady,.. so I don't think that it means it's woke based on that. I haven't played the game, but this sounds like bullshit

Video game expert and developer Yorch Torch Games calls the new James Bond 007 First Light "woke trash" and elaborates on his points: "007 First Light's James Bond is being bossed around by a woman. I've watched the first 13 min. of 'gameplay'. Bond: 'How can I help without knowing what's going on?' "Woman over comms: 'By doing exactly what I tell you.' This is a disaster." Meanwhile, the mainstream gaming media is glazing it. Did feminism destroy this game?