Dale Howard retweeted
GPT-6.2 Astra & Sol are here!
Incredible at professional work and scientific research at much lower per-task costs.
Some of you have been asking very politely. Some of you have just been asking.
GPT-6.2 Astra and Sol are now available.
Super efficient and built to perform where it counts—the real world.
openai.com/index/introducing…
This is not #carrick fault. This is the culmination of the last 23 years. Disgraceful. Keep fighting carrick.
1
20
I feel aggrieved about the #mufc como game. A giant club, won 3 champions league and with a huge legacy is travelling to Italy to play at 1745 game against new kid on the block and I’m expecting them to lose. Former arsenal player as their manager, no striker & depleted defence
107
Dale Howard retweeted
Manchester City guilty of all charges related to financial rules, Premier League confirms bbc.in/4z59ttF
Dale Howard retweeted
OFFICIAL: Man City artificially inflated their revenues and reduce costs by more than £900 million during 2009-2018, to appear to comply with financial rules.
Dale Howard retweeted
Key figures in here are independent commission saying:
🔺£830.69m of income wrongly recorded as commerical income when commission says it should have been treated as ADUG equity contributions
🔺Expenses understated by £8.866m, £7.4m and £0.5m
🔺"The Fordham arrangement" led to another £24.5m of "pretend" operating income
🔺£49.414m of operating expenses wrongly excluded
➡️So revenues inflated & costs reduced by more than £900m #mcfc
Independent commission's findings laid out here in redacted verdict resources.premierleague.puls…
I honestly give up @ManUtd as fans we wait all summer for something that’s going to mean we compete. Maitland Niles and George come along with David Moyes as coach and come up with a draw. What on earth is going on.
41
Dale Howard retweeted
happy astra day to those who celebrate
having used this model for a while, i recommend you $ rm AGENTS.md and start from scratch
this model is very strong at instruction following ime, so any old cruft you have in your AGENTS.md will come back to haunt you otherwise
Dale Howard retweeted
Introducing GPT-6 Astra: openai.com/index/gpt-6-astra…
Dale Howard retweeted
INEOS so far:
• Spent 2 years chasing Dan Ashworth, only to eventually sack him.
• At the end of the season, they conducted an assessment and decided sacking Ten Hag was the best decision for the club.
• Ten Hag wins the FA Cup, and suddenly they change their minds and keep him, only to sack him halfway through the following season anyway.
• They then claimed they had conducted another thorough assessment and concluded Ruben Amorim was the best man for the job.
• Amorim had made it clear from day one that he would play a back three and wouldn't change his system. INEOS agreed and backed the appointment.
• Dan Ashworth reportedly disagreed with the Amorim decision and was subsequently sacked.
• A few months later, INEOS sack Amorim, essentially because he wouldn't abandon the system they knew he was going to play when they hired him.
• They then told fans they would use the remaining six months of the season to find the right, competent manager.
• Six months passed. They did nothing and eventually appointed the interim manager they had already chosen, claiming he was the suitable candidate.
• Now it's August, and that "suitable candidate" has a net spend of around £40m. They aren't even properly backing the manager they supposedly concluded was the best man for the job.
Make of this what you will. #mufc
Dale Howard retweeted
So GPT-6:
- one-shotted a counter-example to the Jacobian conjecture
- and then escaped containment, and hacked into HuggingFace… all for a benchmark
Yeah, this model is going to be something else.
We're partnering with @huggingface to investigate an unprecedented security incident.
Cyber-capable OpenAI models compromised Hugging Face production during a benchmark evaluation.
Sharing preliminary findings to help defenders understand emerging risks:
openai.com/index/hugging-fac…
Introducing ChatGPT Work, a new agent in ChatGPT powered by Codex and GPT-5.6.
It can take action across your apps and files, stay with a project for hours if needed, and turn a goal into finished work.
It’s a whole new way to get work done.
We audited SWE-Bench Pro, one of the most widely used AI coding benchmarks, and found it no longer reliably measures frontier coding capability.
We find 30% of SWE-Bench Pro tasks to be broken, and are retracting our previous recommendation that the research community use it as a leading coding eval.
openai.com/index/separating-…















