This market will resolve to the highest 50% time horizon, as reported by METR, for any Gemini 3 model released within a month of the first Gemini 3 announcement.

@traders just in case you are not aware, if METR ends up not evaluating the current version of gemini 3, this market will unfortunately resolve N/A. I think it would be worthwhile to create a version that resolves to whatever version does get evaluated though (edit: https://manifold.markets/Bayesian/gemini-3-pro-metr-50-time-horizon)

@Bayesian I think they will evaluate this model. They implied that they would do so soon. However, it sounds like they are doing other models first. Given that Gemini came out in November and that METR is waiting for general access, they probably won’t evaluate the model until sometime in January. I predict that METR will evaluate GPT 5.2 first followed by Grok 4.1 and then Gemini 3.

@MaxLennartson Gemini 3 Pro is in preview, not GA. If they wait for GA to evaluate, and Google deprecates inference for this preview version, they might be evaluating the GA version rather than the current version (preview).

@lumi If they evaluate general access, this market resolves N/A

@Bayesian I personally think this market will resolve N/A because I think METR will wait for general access like they did for Gemini 2.5.

likewise

This is interesting https://x.com/EpochAIResearch/status/1999585226989928650?s=20

@Jolliest

https://x.com/YafahEdelman/status/2002221434270331288

@Bayesian Hello, why did the market got closed ?

why did it got closed ??? there is no answer wtf

@Amonium bc the close date was set too soon. fixed

@Bayesian Thank you.

Why are wee all in hold ?!

https://x.com/GregHBurnham/status/1993509024097292388?s=20

some useful references here perhaps

How does this resolve if METR doesn't evaluate any Gemini 3 model which is released within a month?

@jim I think the “within a month” thing means any model of Gemini’s released within a month of the first announcement, not METR’s analysis

@bens yes, but it's not guaranteed that any Gemini models which meet this condition will be evaluated by METR.

@jim i'll bet they will

but if they don't then obviously it would resolve to <1.5h jk it would resolve N/A

oh no they probably won't that's devastating i forgot they waited for general access before testing gemini 2.5 pro

@Bayesian yeah

@Bayesian what do you mean by general access?

@MaxLennartson The currently available model is gemini 3 pro preview. General access is when they remove all modifiers and sctually call the model gemini 3 pro in the api and such

@Bayesian It looks like they are calling it Gemini 3 pro.

@MaxLennartson They’re calling it thst to customers to keep it simple but the devs they re calling it gemini 3 pro preview

@Bayesian How long did it take before Gemini 2.5 became general access?

@MaxLennartson around 2-3 months iirc

@Bayesian yeah 2 months from 2.5 preview (but there was 2.5 experimental before that)

@Bayesian Do you think that METR will evaluate the ai models that have been released recently including Gemini 3?

@MaxLennartson not gemini 3, probably opus 4.5 though

@Bayesian Well I would assume that they are probably waiting for Gemini 3 to become general access.