Anthropic's Claude models currently dominate HLE leaderboards with scores up to 65% as of October 2026, establishing a high bar that Google must clear for Gemini to claim the year's top mark on this 2,500-question expert benchmark. Gemini 4 Argon sits at 57.1% on main evaluations while Gemini 3.8 Flash leads a verified subset at 54.9%, reflecting Google's competitive but trailing position amid rapid frontier progress from 45% ranges earlier in the year. Key catalysts include upcoming Gemini releases, DeepMind announcements, and any tool-assisted or high-reasoning variants that could close gaps before year-end resolution. Trader focus centers on release timing, benchmark contamination controls, and whether Google's multimodal scaling sustains the catch-up pace against Anthropic and OpenAI.