The coin flip game does not work. I tossed 2 coins and it broke after that. You cannot progress forward.
Not a great start for "a generational leap in model effectiveness"
What it feels like to work with Mythos
281–290 of 337 posts
Re: What it feels like to work with Mythos
#282Judging by the benchmarks on Artificial Analysis, "a very real leap over every model" is 2-3 points over competitors (say, 62 for Fable 5 vs. 59 for ChatGPT 5.5 xhigh for coding).
Re: What it feels like to work with Mythos
#283What I find fascinating that there is so little substance in this article about the quality of produced code and the medium. Is the code documented and tested? Is it understandable and extendable? Is it secure? What language, framework, database was used? Author mentions judgement and taste - well, is the code tasteful? Will the model rearchitecture the entire thing if I ask it to add new functionality, spending anot…
There also isn't any meaningful articulation of why this is a "leap forward"... literally everything claimed in the article has been claimed in the same breathless tones in articles written a year prior.
I get that there's little sense in arguing with the MBA hivemind, but... c'mon.
I manage two teams of highly motivated, largely pro-AI engineers. Both teams have independently concluded that they needed to ramp down GenAI usage because of code quality / maintainability concerns. Both teams have suffered from protracted outages caused by LLM jank not being sufficiently fenced off and guarded against. Both teams have expressed concern that the code generated by LLMs is far too verbose, full of slop, and rapidly becomes an unmaintainable mess.
These are teams that are building non-trivial LLM solutions (deep agentic data synthesis and multi-modal data tagging). They are using the technology creatively and pro-actively, not just vibe-coding slop and throwing their hands up when it fails. Both teams will continue using GenAI coding agents, don't get me wrong - but the gains are incremental, not transformative, and need careful fencing to make sustainable.
Nothing in these articles resonates as real. People who work in reality don't agree. I don't understand why this shit keeps getting attention (or rather I do, but the reasons aren't good).
Re: What it feels like to work with Mythos
#284Been working on my pet project today with Fable; it seems pretty solid but not too far removed from 4.8; same hallucinating, same type of bugs, same focus in large projects on just doing what you ask and just ignoring whatever that may touch/break/influence. Running tests in the beginning but when fuller context, just 'will run later' and never doing it in the end unless you tell it to (using some assorted swear words). I will keep using it but it's incremental as far as i'm seeing, not the OMG OMG OMG Mythos is here!
Re: What it feels like to work with Mythos
#285Earlier quoted context omitted.
I’ve seen this pattern again and again, and I don’t bother replying. There’s also the “strong statement, and when you contradict it, they point out some particular circumstances that no one cares about”.
I think a lot of us have stopped talking to each other about this. I see it the other way round to you. I see constant scepticism and doubt that LLMs can build anything useful, and whenever provided with examples, the goalposts just move. And at my own firm, I think every developer is generating most of their code using agentic coding. We're still sceptical enough that we are doing the usual heavy handed human review…
It is all the same narratives from around the invention of the power loom if you look into it.
What I take from that time also is that the hand loom weavers were not incorrect. The power loom did not do as good of a job as they did by hand.
You can still by a hand woven shirt today at a premium price.
There is a category error as if quality is the product as opposed to one input of the product.
You probably don't get to be a master craftsman without that quality mindset so they aren't wrong but missing the forest for the trees.
Re: What it feels like to work with Mythos
#286oh look, more overhyped drivel from a non-technical person.
Re: What it feels like to work with Mythos
#287Been working on my pet project today with Fable; it seems pretty solid but not too far removed from 4.8; same hallucinating, same type of bugs, same focus in large projects on just doing what you ask and just ignoring whatever that may touch/break/influence. Running tests in the beginning but when fuller context, just 'will run later' and never doing it in the end unless you tell it to (using some assorted swear word…
It clearly saw things immediately that 4.8 had missed on my projects. But shortly thereafter, having step functioned past those issues and impressed the crap out of me doing so, it got stuck in the usual endless loop talking about stuff more than doing stuff, occasionally deciding to pause so I'd have to whack it to get it going again.
So nope, not the AGI. But definitely an improvement.
Re: What it feels like to work with Mythos
#288Earlier quoted context omitted.
It just got released, it shouldn't matter. We know this model will be cheaper and faster with time. And we have not even reached the timespan/timeframe were we have ASIC style models. OpenAI has to do something which will beat Fable otherwise Anthropic won. China currently overtakes cars, pv, batteries and very soon silicon chip making, it has all the incentive to also take over AI.
> We know this model will be cheaper and faster with time Why? Demand for AI compute seems to be increasing faster than new production is due to come online for the foreseeable future, particularly if more-intensive models induce demand.
The cost for a fixed level of intelligence will fall over time. We've already seen this. But the cost of cutting edge models will probably not fall anytime soon, and may increase.
So I would expect Fable-level intelligence to get cheaper.
Re: What it feels like to work with Mythos
#289Earlier quoted context omitted.
now for the best question: whats your ROI here?
Humans are very expensive, so the equation almost always falls against them. It's not just salary, but also safety/labor regulation, legal risk, vacations, sick time, personal conflicts, HR, benefits. Even when automation is more expensive on paper, it's generally still cheaper
The real issue IMO is what Richard Sutton alluded to in his talk about GenAI creativity. These models are quite creative at problem solving, but for now, they still need a human to set direction and to occasionally pull them out of doom spirals of churn.
However, given this model now silently corrupts its own work if it thinks you are up to no good, it's absolutely 100% not Mythos so possibly Mythos is better, but who knows now that the alignment and safety safety people are on the case, inadvertently keeping humans in the loop?
https://simonwillison.net/2026/Jun/10/if-claude-fable-stops-...
Re: What it feels like to work with Mythos
#290Earlier quoted context omitted.
> Does AI make incredibly inefficient code most of the time? Yup. But it does it at lightspeed with minimal effort. This hits the nail in the head. Detractors often hang on to examples of coding assistants making mistakes or output subpar code, but they somehow miss the fact that coding assistants can also be prompted again and refactor whole swaths of code just as fast as they introduce oopsies. This means that the…
I think this is overlooking the fact that assigning a coding assistant to fix the bugs it re-introduces for all eternity just leads to spiraling token costs, which might cost more than just hiring a competent engineer in the first place.
This has been a debate for ever, long before LLMs. On the one hand you have people who don't care, on the other you have people who produce good code.
Doesn't matter how fast you can make the wrong thing.