Controversies on model nerfing aside, Fable seeing its biggest performance gains on more complex tasks is a clear indicator of large AI capabilities jumps ahead.
(We’re far from saturating the frontier and far from defining / measuring the next wave of frontier tasks)