Leaked Gemini 4 Neon benchmarks show sweeps across 17 evals vs Claude, GPT
nevrekaraishwa2 · x · 2026-11-19
An unverified benchmark sheet for "Gemini 4 Neon" shows it beating Claude Opus 5.5 and GPT-6.2 Sol on most of 17 evals: GraphWalks long-context 95.2% vs 66.8%, SWE-sweep provider harness 39.0%, SEC-bench Pro 90.5%, WebExploitBench 51.3%, AutomationBench 56.3%. Claude leads only Terminal-bench 4.1 (65.8%) and FrontierCode 2.0 (47.3%); GPT-6.2 Sol tops Chartography at 75.1%. If real, Google's next model pulls far ahead in agentic coding, cybersecurity and long context.
More from Models
- Kimi K3.1 Pro beats K3 on 14 of 17 benchmarks; pairing with Opus 5.5 halves spend — Kimi_Moonshot · 2026-11-19
- Simular box hits OSWorld v2 SOTA at 80.7% success, $1.79 per task — 8x cheaper than August — xwang_lk · 2026-11-19
- Peking University Releases Foundation Model for Measuring and Predicting Human Behaviour — PekingUniversity · 2026-11-19
- Heretic, a 42k-star open-source tool, cuts Gemma 3 12B refusal rate from 97% to 3% — aigclink · 2026-11-19
- Jeremy Howard reverts Solveit models to pre-6.2 versions over quality issues — jeremyphoward · 2026-11-19
- Google's $100/mo AI Ultra plan bundles perks worth nearly the entire subscription — DynamicWebPaige · 2026-11-19