🏅 Top traders

resolving to 16–18 hours because the model's 50% time horizon was measured at 17 hours 25 minutes. considered other possibilities carefully but this is by far the cleanest.

@jim Didn't they only announce that it was at least 16 hours, without providing an exact estimate? Feels a bit early for a resolution when METR themselves seem uncertain

@AlvinAnestrand they provided an exact estimate of just under 17 hours and 25 minutes

@jim any closer to a resolution? I have a lot of mana tied into here sooo

@prismatic all right gimme 10 mins. probably just resolving to 16-18 hrs

I've closed this market for now and will consider how best to approach resolving it.

@jim I think the market should stay open for the time being.

I would suggest closing this market sometime next week to give traders time to catch up with the news and then resolving to whatever the number is under the next version of their methodology, since clearly the current one is not equipped to handle most of this question's options to begin with.

@jim if you do this, I would prefer you N/A the market

Could this market not get resolved yet? On the METR graph they say early next to the models name which implies that this is not their final estimate.

Mythos broke the METR graph, but they say "Likely at least 16 hours" and in their dynamic plot they say 17hrs

This seems very low to me? 43% of 30-50 puts that at 4% per 2h increment, on average. Meanwhile, converting from AECI puts Mythos at around 40 hours from my calculation--so I'd have the mode much higher. AECI is strictly a superset of publicly available metrics right so it's our most trusted source as I see it.

I know METR will be degenerate (massively noisy) at 40h, but is there reason to expect it to be downward biased?

@JacobPfau https://x.com/AlexBarry4/status/2046951516754280600?s=20
24-64h roughly

@JacobPfau I was expecting a 50% reliability time horizon of 40.8 hours and a 80% reliability time horizon of 5.5 hours.

Are we expecting metr to evaluate the Claude mythos preview that was released today?

@MaxLennartson I'm not expecting it. Maybe they'll eval a later version? Otherwise market will resolve N/A if no qualifying model is released so that METR can evaluate it by September 30th

Did some careful studying of relevant trends, think there's a decent chance of >30h time horizons

You should add an option for "Nothing named Claude Mythos is released in 2026"

@Mira resolves N/A if no qualifying model is released by September 30th, 2026