Google’s recent Gemini releases, including the October 13 launch of flagship Gemini 4 Argon and prior 3.x Pro variants, have emphasized competitive positioning on expert benchmarks like Humanity’s Last Exam (HLE), a 2,500-question multi-modal test of frontier academic reasoning created by the Center for AI Safety and Scale AI. Earlier Gemini 3 Pro scored a leading 37.5% on HLE in late 2025, but current leaderboards show Anthropic Claude Opus 5.5 and Fable 5.1 variants ahead at 61–65%, with Google’s 3.8 Flash and 4 Argon models trailing in the mid-40s to mid-50s range depending on the variant and tool use. Delays in planned 3.5 Pro releases, DeepMind leadership shifts, and rapid Flash-tier updates highlight Google’s focus on efficiency and integration over immediate HLE dominance. Traders watch for official model cards or announcements tying the next Pro iteration to new HLE results, as labs increasingly cite the benchmark in launches amid tightening competitive gaps.