This will be evaluated according to the AI Safety Levels (ASL) v1.0 standard defined by Anthropic here. See this market for criteria for determining a system to be ASL-3 for the purposes of this market.

Unlike in that market, the date in question is the date of the first public report that contains credible evidence that a model is ASL-3, which may be later than the date that the model is trained, and earlier than the date that there is a consensus that that evidence should count for ASL-3.

Feel free to add new answer choices. Valid choices must be in the format YYYY QQ.