🤣🤣🤣

On the new ChatGPT Pro $200 plan. Updated limits, not the old 20x. Had two pretty different days on it, so thought the usage was worth sharing. Yesterday was mostly GPT-6.1 Sol Medium, generally 1 to 3 projects. Around 8 hours and I'd only used roughly 10%. About 1.3% an hour, so very roughly 75 to 85 hours if you kept that pace. Today I pushed it harder. Mostly Ultra, max reasoning, 2 to 6 projects, plus multi-agent reviews and validation work. After about 10 hours I'd used roughly 51%. About 5% an hour, so closer to 20 hours at that intensity. Roughly 4x the burn between the two days. That doesn't mean Ultra itself just uses 4x more. Today was a heavier workload, with more projects, more agents and more reasoning going at the same time. Medium + fairly normal concurrent work: around 75 to 85 hours. Ultra + heavy multi-agent/concurrent work: around 20 hours. I'd expect a single project to stretch further. The more agents and reasoning you throw at it, the quicker the allowance goes. Personally I think that's fairly reasonable for the money. I've used plenty of models over the past few months where usage disappeared a lot quicker for the amount of useful work I actually got back. I still probably wouldn't sit on the $200 plan all the time. For me it makes more sense to move up when I know the extra allowance is actually going to get used. Just what I've seen from my own use so far. Yours could look quite different depending on how you're using Codex.

On the new ChatGPT Pro $200 plan, so the updated limits, not the old 20x. About 8 hours of non-stop workflows, bouncing between 1 project and 2 or 3, and I’m on about 90% remaining. Mostly GPT-6.1 Sol Medium, with a couple of Ultra runs. Rough maths puts the full allowance around 80 hours. Can’t complain, depending what you’re actually doing.

⚠️ A whistblower who works as a border control officer at Manchester Airport has come forward. He says there is a private terminal being used to fly in migrants. "We patrol to check nothing is untoward. There is an area we can see - but are formally prevented from entering or inspecting. Regular flights arrive there - 3 or 4 times every week. Transponders are switched off, yet they are given clearance to land using a special call sign. Passengers are not screened by us - they are processed directly by Home Office personnel. They board unmarked buses. Each flight carries 200 - 300 people. These flights arrive from the Middle East and sub‑Saharan Africa. They land between 2:00 am and 4:00 am. Government gives clearance. Unmarked buses collect passengers and leave through a hidden exit gate onto restricted roads - then merge onto motorways and vanish". Full interview: jaoc.org.uk/whistleblower-in…

Renewal time. ChatGPT or move to Claude. I want more usage this month, so I’m looking at a $200 plan either way. OpenAI just cut the allowance on the $200 tier, and Sol 6.1 has been sitting around 20 tok/s from what some people are seeing. Tibo says it’ll speed up. He has also said Codex on the 5x plan really does see about 5x the weekly usage. Claude’s 20x is the same price right now, but the labelling is the bit people seem to have forgotten. Max is sold as 5x or 20x Pro usage per 5-hour window. The weekly cap on the $200 plan looks closer to about twice the $100 plan, not four times. Past reports put the actual multipliers well under the marketed 5x and 20x. OpenAI has shipped some decent stuff, Dots and the rest, but none of that is available to me in the UK yet. Also not sure how much Astra you’d actually get on the $200 plan before it rinses the allowance. Leaning toward one $200 plan and the other vendor on $20.

Morning Campers Diesel at a record high £2 a litre It’s £130 quid to fill my tank So filled her up and drove off £75 court fine That’s a saving of £55 For more money saving tips follow me 😂

😧

Pro $100, GPT-6 Sol Ultra: 9–10 hours of continuous work and only about 25% of the usage gone. Fine for Sol and Luna. Astra still eats the plan alive.

For AI access don’t think I could justify today anything more than a 200 plan otherwise I’d start considering investing heavily in some better local hosting tbh for personal stuff..

Are OpenAI always playing catch-up? That’s the feeling at this point. Model intelligence, model releases, now a Grok Bot competitor as well. Everything I’m seeing still screams rush to market. DevDay yesterday was huge on paper. Dots, GPT-6.1 Sol, a pile of Codex stuff, twenty-odd announcements. Some of it isn’t even live yet, and some of it isn’t available in Europe or the UK at all. So you get the keynote and then… wait. Anthropic put out Opus 5.5 and then Sonnet 5.5 and suddenly OpenAI look like they’re trying to do everything at once just to keep pace. Biting off more than they can chew, if I’m honest.

Codex banked reset just dropped

someone take City out of the oven they’re done

Little summary

Tbh im ready for Grok 4.8 and it best be good.

Liftoff of Starship's first orbital flight

Danny Tommo has been remanded in custody for popping an empty boat But suspected terrorists with a bomb going to attack an RAF base have been released on bail WTF!!!?????

My 9 games 🤩😅

🇬🇧 💀 Gonna go broke before I ship 😂

BREAKING: Diesel is selling at its highest price ever recorded news.sky.com/story/money-liv…