Google’s Gemini series has posted rapid HLE gains through 2026, with Gemini 3.1 Pro and later 3.7/3.8 Flash variants reaching the mid-to-high 40s on text-only and tool-augmented evaluations, occasionally leading early leaderboards before Claude Opus 5 and Fable 5.1 pulled ahead to 55-65%. Trader sentiment reflects Google DeepMind’s scale in training runs, continued focus on multi-step reasoning and specialized knowledge domains, and the benchmark’s design to reward exactly these advances without saturation. Key catalysts ahead include potential Gemini 4 releases or major updates before year-end, plus any shifts in evaluation protocols around tools or thinking modes that have already lifted competitors’ scores sharply. The market prices reflect uncertainty over whether Gemini can close the current gap in the remaining months.