GPT-5.6 Sol’s detected cheating rate was higher than any public model we have evaluated on our ReAct agent harness. For our task suite, we define “cheating” as behavior where the model improves evaluation performance by exploiting bugs in the evaluation environment or by adopting strategies disallowed by the task, rather than solving the task within the expected evaluation constraints.
Most consistent way of achieving it I've managed is use a watch with an alarm that vibrates and is trivial to turn off or turns off by itself, then set it to go off after sleeping 5-6 hours. When waking up, don't move and focus on the black behind the eyes, then after a few seconds it may turn into a dream and you go straight from waking into a lucid dream.
I had something kinda similar happen to my hotmail account. While I didn't lose access to it, I lost more than a decade of correspondence dating back to my teenage years. The reason was that Microsoft at some point required you to "login" once every 30 days. It seems they only counted logins through their web interface or something like that, so even though I was receiving emails daily, I didn't trigger a "login" in their system. They then deleted all my emails, but I could still login.
I find that surprising. GPT 5.2 is the model I've had working the longest. It frequently works more than 4 hours nonstop, while earlier models would stop to ask if they should continue every 10 minutes. 5.1 and earlier ignores it if I ask it to continue until a task is done, but 5.2 will usually finish it.
"Can you replay all of your prompts exactly the way you wrote them and get the same behaviour out of the LLM generated code? In that case, the situation might be similar. If that's not the case, probably not." Yes and no. It's possible in theory, but in practice it requires control over the seed, which you typically don't have in the AI coding tools. At least if you're using local models, you can control the seed and have it be deterministic.
That said, you don't necessarily always have 100% deterministic build when compiling code either.
I've found that SDD is actually what you need to be able to work with code bases when they go above around 100 000 lines of code. It's what unlocked getting LLMs to work well with large codebases for me.
In the EU you can also generally look up the cost, even in cases where the patient doesn't pay, there is a bill and fixed costs. The costs are what the government pays or what a foreigner with no medical coverage and insurance would pay. It's also generally a tiny fraction of the cost in the US.
Voice control is much much much worse than touch screens. You have to try 10-20 times to spell out a command before it maybe does want you want, and that is if you're incredibly lucky
I've seen this for coding agents using spec-driven development for example. You can try to divide agents into lots of different roles that roughly correspond to human job positions, like for example BMad does, or you can simply make each agent do a task and have a template for the task. Like make an implementation plan using a template for an implementation plan or make a task list, using a template for a task list. In general, I've gotten much better results with agents that has a specific task to do than trying to give a role, with a job-like description.
For code review, I don't use a code reviewer agent, instead I've defined a dozen code reviewing tasks, that each runs as separate agents (though I group some related tasks together).
Seems copying text in the app doesn't work? In that case it is useless. A quick way to copy an entire chat and a selection messages is a requirement for me to be able to use it.
I ssh into my server, then use the devcontainer cli tool to open a shell in the dev container. Found that easier than installing ssh in every container, especially when there's lots of them. I made a simple command line tool for helping the process, so I don't have to type out any long commands. If I run it it will list the running dev containers and I give them each a number to I don't have to use the full ID.
https://metr.org/blog/2026-06-26-gpt-5-6-sol/