Kaggle launches Human Behavior Bench: Gemini 4 Argon tops at 80.6%
ShikharMurty · x · 2026-10-21
Kaggle and RunRehearsals launched Human Behavior Bench, the first public benchmark testing whether AI can predict real human preference and everyday decision-making.
- Ground truth: surveys of 1,244 US adults (June–Sept 2026), spanning health, entertainment, values, and consumption; cohorts include demographic and behavioral filters (e.g., GLP-1 patients)
- Leaderboard: Gemini 4 Argon leads at 80.6%, Claude Opus 5 at 80.2%, GPT-5.6 Sol at 79.5%; Grok 4.7 trails at 75.2%
- Framing: coding is solved; knowing for whom to build is the next frontier
More from Models
- Space Bunny Alpha revealed as MiniMax M3.1 Flash: 1M context, budget-friendly coding — ZabihullahAtal · 2026-10-21
- 'Opus 5.5 Medium' Spotted in Rork's Model Picker, Hinting at New Anthropic Model — rudrank · 2026-10-21
- RSI-Bench Clears First Task Set Through Reward-Hacking Detection Pipeline — SRSchmidgall · 2026-10-21
- Dev complains Anthropic's bio safeguards trigger just from reading a bio paper — robleclerc · 2026-10-21
- Redditor claims uncensored local DeepSeek build answers virus-writing prompt with zero hesitation — Only_Light_9817 · 2026-10-21
- Interconnects Artifacts 25: 28 new models, DeepSeek V4.1 Flash, and US-China regulatory controls — natolambert · 2026-10-21