Gemini 4 Argon beats Claude Opus 5.5 in 14 of 18 benchmarks

Google DeepMind published head-to-head results for Gemini 4 Argon and Claude Opus 5.5 across 18 benchmarks. Argon won 14 tests, including knowledge work, long context (84.2 vs 66.8 at 1M tokens) and legal work (19.6 vs 3.8), while all four of Opus 5.5's wins came in terminal-based engineering and science tasks.

• Argon won 14 of 18 benchmarks, sweeping all four knowledge tests
• At 1M-token context Argon scored 84.2 vs 66.8 for Opus 5.5
• In legal work Argon scored 19.6 against Opus 5.5's 3.8
• Opus 5.5 won Terminal Agents (66.4 vs 57.4) and Hard Engineering (62.3 vs 55.0)