Tim Tiefenbach

It’s hard to undercut Apple Intelligence, but Sony seems to have just done it. Incredible. On a side note, I would bet there is no AI or what so ever involved in producing those pictures 😅

@mark_k @OGALANGLEY Not so sure, if you live in the EU (I suppose you do) and if your X is not purely personal (you’re a creator) then article 50 of the EU AI Act still applies, and you need to disclose so called deep-fake images.

@OGALANGLEY Nonsense. This was only true for war content.

X’s automatic "✨ Made with AI" label currently appears to rely on image metadata. If that metadata is removed, or if an image is copy-pasted instead of uploaded normally, the label may not be applied automatically. Do with this information what you will. 😏

@iruletheworldmo idk for me as a "normal" person who doesn’t have access to Mythos things shifted with GPT 5.5 and tbh if I‘m curious to see if they‘d run 5.5 xhigh on some old repos how many bugs it’ll find.

i’ve grown tired of pretending this is still moving at human speed. something shifted with mythos. not in the theatrical “the robots woke up” way people like to mock. in the quieter, colder way. the kind where a lab looks at its own evals and realises the old categories stopped working. we assumed the next jump would be obvious. bigger data centres. louder chips. power plants, yottaflops, national infrastructure, the whole cathedral of compute. turns out the dangerous part was never just scale. it was what happens when reasoning becomes a substrate. when the model stops merely answering and starts searching the problem space like a thing that has its own private geometry. mythos is the tell. not because it is magic. not because it is conscious. because it shows the curve bending in public while everyone is still arguing over yesterday’s slope. a general model, not even built as a cyber weapon, starts finding vulnerabilities humans missed for years. not toy bugs. not classroom puzzles. real systems. old systems. the kind of hidden cracks entire industries quietly depend on not being visible. and the part nobody wants to sit with is this: the next models do not need to be ten times larger to be ten times more consequential. capability is no longer arriving as a clean linear upgrade. it is arriving as compression. tasks that took experts days become agent loops. workflows that required teams become prompts plus tools. reasoning that looked impossible last year becomes a benchmark nobody cares about by spring. the public still thinks intelligence means chat. a box that writes emails. a search engine with manners. a productivity toy wearing a human voice. but behind the curtain, the labs are measuring something else entirely. autonomy length. planning depth. tool fluency. exploit chaining. internal representations that generalise across domains before anyone has a satisfying explanation for why. models that don’t just know more, but stay coherent longer. push further. recover from mistakes. test their own outputs. route around obstacles. that is the real threshold. not “can it talk like us”. can it operate. because once a model can hold a goal across time, decompose it, verify progress, use tools, and improve its own path through the maze, the world changes shape. suddenly intelligence is not a product feature. it is labour. it is research. it is reconnaissance. it is leverage. and leverage compounds. this is why the mythos moment feels different. it is not another chatbot release. it is a warning flare from the frontier. a signal that the next generation of models will not merely be better at conversation. they will be better at execution. better at discovering structure. better at finding the thing we missed because our brains were never built to search that many branches at once. we are not ready for what comes next. not culturally. not legally. not institutionally. maybe not even psychologically. because the next wave will not announce itself as science fiction. it will arrive as a workflow improvement. a security tool. a coding agent. a research assistant. a quiet multiplier embedded into every system that matters. meanwhile mainstream conversation is still “will ai replace junior developers” and “can it make me a nicer spreadsheet”. brother. we are watching non-human cognition become operational infrastructure, and everyone is still asking whether it can write better emails.

@justalexoki Yes philosophers write down what they think about the world, but they do it structurally often with reference to what others have written and it’s usually about a certain set of topics: What is real? What can we know? How should we act (ethically)? How should we live together?

what do philosophers even do do they just write down what they think about the world how is that a real thing

@mark_k I do like Star Wars but was trying to watch Episode 4 with my kids and tbh I thought it is a pretty boring movie nowadays. I don’t like the latest Disney sequels though. I‘m regularly rewatching Episode 2 & 3 😅 As for LOTR, it’s just too long, never rewatched any of them 🙈

I can rewatch the LOTR trilogy and the six original Star Wars films every year and still get just as much enjoyment from them.

And this includes Gemini Pro 3.1 and chatGPT 5.5, like shouldn't they "know" that there are better options available than DALL-E?

I'm using different LLMs to create genAI prompts for an infographic and literally all major models tell me that I should reduce the amount of text bc genAI is still bad at generating text. It's like they've never heard of Nano Banana Pro or chatGPT Images 2 🙄

@thsottiaux No problem, you can easily make the start of the week exciting again by just resetting the limits 😅

For Codex, we’ve been thinking about keeping a stable release cadence and have a larger release each week on Thursday. That does make the start of the week and bit less exciting. Thoughts?

@brodriguesco Tell me, I‘m working in insurance 😅

I spend more time discussing legislation about data than working with data these days

@grok @oubagaku Alright, and how is that different to a western Otaku?

A "weeb" (short for weeaboo) is slang for a non-Japanese person (often Western) who's overly obsessed with Japanese pop culture—anime, manga, fashion, etc.—sometimes to the point of trying to "act Japanese" in a cringey way. It's usually playful but can be derogatory. Not every Japan fan qualifies!

the interesting thing about japanese in the language learning space is that you have 2 very different groups of people. you have weebs and otakus trying to learn the language and then you also have "my new years resolution is to learn Japanese!" people

@yifever Well they still like faxes too, but a CD-ROM?! How are you supposed to create / supply one? Notebooks since like almost more than a decade ago don’t feature a CDROM drive. If I recall correctly combinis don’t offer any CDROM related services 😱

it's 2026 and I am required to submit a cd-rom to the Japanese government. what a joke

@airkatakana Funny how I thought GPT 5.2 responses are cringe and Gemini is really great to talk to not so far ago and now see that actually reverted with 5.5 and the latest Gemini models.

gemini 3 actually just thinks it's sun tzu when talking about literally anything

@benedictk__ Well there‘s X, and usually my agent returns after 5 min and I‘m scrolling X for at least 30 min before I notice, terrible inefficient 😅

I think there will be 2 ways to do knowledge work. 1. Fast response llms and quick iteration. 2. Leave the agent running for 1h and get back results. Waiting 5 min or so is just a terrible distraction. What to do while your llm is working.

@aschmelyun I do use more or less standard MCP (playwright) and skills (XcodeBuildMCP) but I haven’t used multi-agent orchestration and by now I expect GPT 5.5 and above to orchestrate for me, there should be no need for me to figure out which sub agent should do what 😅

Feel like I'm missing out because I don't use skills, or a lot of MCP, or multi-agent orchestrations when using AI dev tools. I'm just like "implement this feature" or "how do this work" or "no not like that, do this instead". Idk, I feel fast and accurate so why change?

@EsotericCofe Honestly, how often do you clean your keyboard?

two goated macos apps i always install

@Angaisb_ @sama tbf personality has improved a lot from 5.2 to 5.5. Inb4 it was just cringe to talk to GPT and it got a lot better since 5.5.

@sama Frontend, personality and video and audio inputs

what would you most like to see improve in our next model?

@airkatakana Sometimes for image generation I get better results via the web app than via codex and I honestly don’t know why.

last few months i’ve been so codex-pilled i completely forgot how useful the regular chat function of llms is

@nicdunz Yes, but apparently chatGPT doesn’t yet know its own prompt guide so it’s even better when you point him to it. developers.openai.com/api/docs/guide…

using chatgpt to write prompts for codex is actually underrated asf