Mike Ross

@m_franceschetti Thanks for everything you do! I've been a happy customer for many years now and look forward to even more innovative sleep solutions from Eight Sleep in the future.

@matt_elevenlabs 11B (and congratulations)!

We just raised $500M at an 11B valuation 🎉 To celebrate, we’re giving away 1,000 free credits so you can test our platform. For the next 6 hours, comment “11B” below and we’ll DM you the credits (must follow) 👇

@elonmusk @MKBHD I respect that Tesla has to make the best business decision for itself, but I'm pretty disappointed that you are discontinuing your larger and more upscale models.

@MKBHD New Roadster will be incredible

Oh Roadster is so cooked

@niccruzpatane They should name the baby Nikola in tribute?

Tesla Full Self-Driving drove his wife to the hospital while she was in labour. Autonomy is here, and it’s amazing.

@DeryaTR_ @biogerontology @InSilicoMeds Thanks for everything scientists like you do, Derya! You're humanity's true rock stars and your collective work is building a better and more prosperous future for us all.

It was great meeting my friend @biogerontology, the visionary pioneer founder of @InSilicoMeds & behind the first AI-discovered drug, in clinical trials for pulmonary fibrosis. He is now developing dozens of new treatments using AI. We also talked about our plan to reverse aging!

@pbeisel Awesome article as usual Phil! Quick question: Does the Tesla action model utilize weather data as an input token (eg below freezing or not)?

@davidasinclair Eyebrows are not the only thing that Viagra is raising. All jokes aside, interesting finding.

Low dose viagra to counter the effects of cardiovascular & brain aging raises eyebrows but it’s a real thing

@DevinOlsenn Makes me wonder whether Tesla is using the interior camera and/or microphone to determine when you are finished with your order. You could certainly train a model to recognize which patterns in the cabin correspond to the vehicle moving on from the order/pickup windows.

FSD V14.1.1 Drive Thru V2 Last night I posted a video of FSD V14 handling a drive through perfectly. I rolled up the window after each 'stop' and held my hand out of the window at the end to try and keep the car from moving. Well this time I did none of that, and the car actually did better than last time. Incredible stuff - how do you all think @Tesla_AI has managed to do this?

@sama Definitely smarter as I can effectively mitigate slower response speeds by utilizing asynchronous workflows and parallel prompting on my end. That said, much smarter and a bit faster is probably the sweet spot for most people.

thanks andrej! do you care more about it getting smarter or faster?

@aphysicist "We wanted flying cars, instead we got 140 characters."

today i ran an experiment that exposed what i believe to be one of the biggest threats to america right now i attempted to raise funding for a manufacturing company from silicon valley VCs and from PE funds the VC gave me 5 minutes on zoom and said this was "more of a PE deal" because it was "too capital intensive", and they had no comps to benchmark it to, so they left to hop on another zoom and give an AI saas company a $50M pre seed the PE fund had no one to answer. i tried two others same thing. finally got a hold of the fourth which said they "don't fund r&d" and "we buy ebitda", and suggested i talk to VCs about de-risking the tech first so i call back the VC and ask "what about tesla or spacex?" and they said "those aren't replicable, you're not elon" while funding the 47th AI email assistant this is the reality: america is being priced out of its own industrial future while companies get stuck in the valley of death between VC and PE this is why 50% of our founders don't care anymore about building hard tech / manufacturing, because VCs want guilt free 100x returns from the saas casino, and pe wants existing cash flows if we want to support reindustrialization and domestic manufacturing, we need investors who understand the middle ground exists, and real tech can enable massive empires otherwise the factories of tomorrow won't be american, they'll be chinese. but at least we'll have another b2b saas platform with a 40x revenue multiple

@DirtyTesLa The Bermuda Triangle for autonomous vehicles.

What insane driving obstacles are in this area? 😱

@aubreydegrey Incredibly grateful for the work pioneers like you have contributed to the field, especially given the career risks you must’ve faced early on.

It's gratifying to see more and more credentialed experts publicly embracing the LEV concept. For far too long, virtually none of my colleagues would do so. Now let's get there.

@testingcatalog @sama I have access to GPT-5 Pro on Teams now, but I still don't have any way to turn on the legacy models. I've tried both the regular and Workplace settings multiple times. Have any Team subscribers been able to get this functionality to work?

@sama Any updates on GPT-5 Pro for team accounts? The previous o3-Pro is no longer available there 👀

today we are significantly increasing rate limits for reasoning for chatgpt plus users, and all model-class limits will shortly be higher than they were before gpt-5. we will also shortly make a UI change to indicate which model is working.

@testingcatalog @OpenAI The lack of transparency in what paid users get is utterly ridiculous. If @OpenAI is going to have rate limits, then they need to clearly post them! Also, what is the difference between using GPT 5 Thinking and the Think Longer setting with GPT 5? How does it affect the limits?

@OpenAI Will Pro be available on Team accounts too? 👀

A few GPT-5 updates heading into the weekend: - Now rolled out to 100% of Plus, Pro, Team, and Free users - We’ve finished implementing 2x rate limits for Plus and Team for the weekend; next week we’re rolling out mini versions of GPT-5 & GPT-5 thinking to take over until your limits reset - GPT-5 thinking and GPT-5 pro now in main model picker - GPT-4o is now also available to Plus and Team users. To use it across platforms, go to settings on ChatGPT web and toggle on “show legacy models.”

@OpenAI Why am I paying more for Teams if I don't get access to the legacy models like the cheaper Plus accounts do? Additionally, why don't you clearly publish what the rate limits are for each tier? I'm sick of having to guess or look for the answers on X.

@bryan_johnson Out of curiosity, do you think your body might be less resilient to stressors now due to the fact that it's used to an absolutely optimal regimen? I wonder whether an average person would experience the same degree of deterioration. Regardless, great job - you're an inspiration

Travelled internationally and went really hard for 11 days. As an experiment, made the trade off to accept less sleep and lower recovery. Was remarkable to experience how fast my body and mind deteriorated from peak ability and experience.

@emollick I'm not sure why anyone would actually value disfluencies in their AI's speech. I'm looking for the smartest and most efficient communication in my assistant, not a gimmick that mimics an annoying intellectual limitation of human beings for the sake of familiarity.

The new ChatGPT Advanced Voice Mode is super interesting - lots of deliberate use of disfluencies (nervous laughs, ums & ahs) and vocal changes make it feel much more human than the previous version Really shows the possibilities from multimodal voice vs most AI’s text-to-speech

@TeslaNewswire Just like my Roomba

Tesla CyberCab will feature wireless charging that sets a new benchmark: ✅ Delivers up to 25 kW of charging power ✅ Achieves over 90% efficiency ✅ FSD automatically aligns the vehicle perfectly with the charging pad

@gdb Terrific work, but why is this not being made available to ChatGPT subscribers? To be frank, 4o is not good enough anymore relative to Google's new models and it's getting increasingly difficult to justify not switching.

New model in our API — GPT-4.1. It's great at coding, long context (1 million tokens), and instruction following.

@sama @amasad You should bring back the 'hold the circle to pause model response' feature from the earlier voice UIs. It should work as it does currently unless the user holds the UI circle, in which case it should pause. The current UI forces you to break your train of thought to interrupt it

@amasad not with 4.5, but we want to make it much much better. super important for future products.

trying GPT-4.5 has been much more of a "feel the AGI" moment among high-taste testers than i expected!