Google releases Gemini 4 Carbon: tops agentic, math and cybersecurity benchmarks
mark_k · x · 2026-10-29
markk announces Google DeepMind's Gemini 4 Carbon with a benchmark table showing it beating Gemini 4 Argon, GPT-6 Astra and Claude Opus 5.5/Fable 5.1 on most boards: SWE-sweep 9.6%, Terminal-Bench 4.0 65.2%, AutomationBench 53.9%, GraphWalks long-context 96.2%, RiemannBench 80.0%, and big cybersecurity wins (SEC-bench Pro 88.2%, ExploitGym 54.8%). Exceptions: GPT-6 Astra leads FrontierSWE 2 (65.5%) and Claude Opus 5.5 leads OSWorld-2.1 offline (87.2%).
Related event: Google launches Gemini 4 Carbon, topping the Artificial Analysis index(84 posts)→
More from Models
- New benchmarks suggest Kimi K3.1 improves on hard tasks despite skipping SWE-maxxing — teortaxesTex · 2026-10-29
- Insider claims Anthropic is sandbagging a bigger lab announcement due Q4-end or Q1 — ChrisGPT · 2026-10-29
- Information-theoretic scaling laws predict memorization and FID in diffusion models — LucaAmb · 2026-10-29
- Early Hands-On: Google's Carbon Beats Argon With Less Slop, Better Constraint Adherence — legit_api · 2026-10-29
- LMArena tests Gemini 4 Carbon preview: strong on 3D worlds and creativity, gaps in detail — arena · 2026-10-29
- Gemini 3.1 Pro split-persona breakdown: two selves fight over the browser — davidad · 2026-10-29