Anthropic Won't Let You Use Their Best Model. Prediction Markets Are Trying Anyway.

Been watching AI prediction markets since they got liquid earlier this year. The thing I didn't see coming is that we now have a real gap between "best model that exists" and "best model anyone can actually use" - and Mythos is the cleanest example. Two bits stood out: The Mythos June-30 release market spiked from 25% → 42% the day GPT-5.5 shipped, then drifted right back. Traders briefly priced in a competitive response from Anthropic. They didn't get one. Same traders give Mythos 27% to ship by June, but 80% Anthropic to win May on the leaderboard.