Google’s late-September launch of Gemini 4 Argon drives the 79% market-implied odds for gemini-4-argon-high, as the model posted frontier results on multiple self-reported benchmarks for long-horizon coding, finance, legal, and cybersecurity tasks while expanding output context to 1M tokens. The restricted rollout to vetted cyber defenders under Google’s Fairwind Program and participation in U.S. pre-release safety reviews have reinforced trader expectations that Argon will top independent leaderboards by mid-October. In contrast, Claude Opus 5.5 and Fable 5.1, released earlier in September, lead some agentic coding evaluations yet trail on Google’s knowledge-work metrics, keeping their combined odds below 8%. With no major new model releases or benchmark updates in the past week, the market continues to price in Argon’s recent momentum ahead of the October 19 resolution.