[flagged]
GPT-5.6
871–880 of 1001 posts
Re: GPT-5.6
#872Not specific to OpenAI / Codex, but I'm curious what people are doing to protect themselves from any destructive actions by their coding agents? Just install and pray? Explicity approve all actions? Reconfigure for safety? Run in a sandbox (Docker) ?
YOLO. Btw for real tho, if you don't have the time or means to mess with full sandboxed environments, just working within a git repo and instructing on your agents.md project level that the agent should back up dirty files (local changes that were not yet committed) before changing them is enough and super fast and easy to set up. And by back up I just mean a simple instruction to back up to some temp location under…
I literally have had a SOTA Codex agent delete a bunch of files just last month, but it took very specific circumstances. It was working within my game repo, in a branch, and it ran out of SSD. It had to free up some space to work, so it looked around and found 25GB of untracked files in the project, in a folder called Recordings. Might as well clean that up, it thinks. There goes all the raw Unity Recorder footage I'd ever recorded of my game, over many years of development.
So yeah, it won't go on a spree outside of its lane even with full access, but if you give it a box and tell it to go ham, it's on you to make sure you didn't leave precious unrecoverable assets in that box.
Re: GPT-5.6
#873Earlier quoted context omitted.
I’d argue the opposite. I’ve switched back and forth from one to the other and Opus/Fable has been constantly better than any GPT in my daily work. It’s a bit slower but it does the things right, with as little code as possible, some comments where needed. Codex is faster but you always have to correct it because it got something wrong; it writes tons of code ("let me add a small helper") with obvious comments.
Purely anecdotally the one persistent issue I have with LLMs writing code is that they are absolutely paranoid and add a load of indirection and defensive crap and even if you prompt to avoid that it will often require manual steering to remove the cruft.
I tell it to avoid belts and suspenders, don't leave dysfunctional code in, and fail loud. Seems to change that behavior.
Re: GPT-5.6
#874Earlier quoted context omitted.
They did. I wonder if Anthropic will also be removing the 50% limit. My Fable weekly limit is at 15% used already, 5.6 Sol at 3% used. And this is with the Max 20x plan compared to Codex 5x. I don't work on the same tasks to compare them objectively, but GPT 5.6 on xhigh seems much cheaper. Essentially unlimited usage.
Anthropic really needs to get Opus 5 out ASAP. The gap between Opus 4.8 and Fable is large enough to drive a GPT 5.6 sized bus through. A better Opus would take some of the heat off.
Provided such a Opus 5 performs on par with Fable 5 and GPT 5.6 Sol.
Otherwise I am not interested.
Supposedly Fable 5.1 is in the later stages of the release pipeline, maybe it takes back the crown from OpenAI, who are now rumored to launch GPT 6 in August.
Re: GPT-5.6
#875Earlier quoted context omitted.
Seeing the dramatic differences in scores just going from high to xhigh is just another demonstration of the bitter lesson: Just keep scaling search and learning. We are probably going to need a lot more GPUs.
There goes my plan to buy a PC for the next decade
I actually think we're in a strange situation with AI compute.
Right now, we have models that are statistical models of language, with a world model and reasoning "falling out" of a lot of effort.
It's like we've made something that's a little bit intelligent, and now we're trying to amplify that trick to create something that's quite intelligent. And - don't get me wrong - it works.
But it's also super, super inefficient. We're having machines "think out loud" to compensate for the quality of their thought processes. We elongate the path to make up for the progress made on a given step.
I tink there's probably a much smarter way of doing things that will require qualitative architectural (and quite possibly hardware) innovations. Right now we're on the path to a Dyson sphere: that's probably not going to be necessary once we figure out a smarter way to think.
Re: GPT-5.6
#876Earlier quoted context omitted.
this is so cool: it's playable (even though super boring since there are no enemies) and you can feel that a few iterations would make it very usable. Which model is the best at the moment, for this kind of stuff, in your experience?
I'd say Fable 5: https://senko.net/vibecode-bench/2026/rts-fable-5.html It even has enemies! (I'm not too mad about it not following my instructions because it can be fun to play :) And I generated that from Claude Code on my phone. Sonnet 5 also produced a pretty nice version. You can see all of them here: https://senko.net/vibecode-bench/
> I'm not too mad about it not following my instructions
I'm a bit concerned about this - starting with GPT5, AI labs started doing this 'complete app from a prompt' sizzle demos. When I started working with GPT5 - which was supposed to be a qualitative jump, just like Fable is now, I tried to do a frontend, and discovered that it gave me a CSS-animated purple-blue interface with embossed buttons, gradient backgrounds and dropshadows.
It looked very cool, but it was a bit overwhelming (also broken), and I was really looking for a pedestrian Bootstrap job.
It required not inconsiderable amount of wrangling for GPT5 to stop doing this. So I don't really like the idea that these models have tons of implicit and hidden behavior, to 'soup up' pedestrian prompts.
Re: GPT-5.6
#877Cannot believe I needed a VPN to the US, to open this from Switzerland...
At least give me the article ffs.
Re: GPT-5.6
#878"We've extended usage of Claude Fable" message incoming any day now.
They reset all usage half an hour ago. It's back to 0% per week and session. No specifically Fable related.
Before the reset, a single 5h window would use up 5% of the weekly quota. Now it uses 10%. Thanks a lot, Anthropic.
Re: GPT-5.6
#879I really appreciate the focus on intelligence WITH token efficiency. I'd like to see that become the trend. Smartest per token metrics. Least tokens to accomplish the task above a certain success level. Most of my tasks would benefit from efficiency / token, but switching models constantly, and trying to guess the right model and effort level takes up too much of my processing.
> I really appreciate the focus on intelligence WITH token efficiency that is a polite way of saying "I don't believe AGI is coming anytime soon".
seems more like "this priorisies AGI"
What is AGI to you though?
Re: GPT-5.6
#880I really wish there was just an easy guide on when to use Sol vs Terra vs Luna, and it just moves further into confusing territory when it comes to naming. The naming convention is especially difficult to decipher depending on what your native language is. Of course a latin language speaker might be able to easily determine oh yeah each one is slightly bigger than the other but I still think it borderlines too confus…
It's just how LLMs work - these are three completely separate models trained in parallel, with different numbers of parameters and using different amounts of compute.
They could hide this behind a harness that picks the correct model for you, but devs don't seem to like that.
There's also the 'effort' slider, which I guess how many experts in the MoE are evaluated and how long reasoning chains are allowed to go on, which is the 'smooth' scaling you are thinking of.