Gemini 4 Hydrogen Benchmark Table Leaks: Leads Agentic Coding, Trails Claude on Terminal-Bench
XueFz · x · 2026-11-13
A leaked-style benchmark table compares Gemini 4 Hydrogen against Claude Sonnet 5.5 and GPT-6.1 Terra. Gemini leads on SWE-sweep (13.5% vs 3.1%/5.8%), Agents' Last Exam v1.1 (71.6%), AutomationBench-AA (76.1%), Vending-Bench 2 ($14,549), MKA (53.4%), Chartography (66.6%) and HLE-Diamond (58.8%), but trails Claude on Terminal-Bench 4.1 (57.0% vs 63.9%), OSWorld-2.1 (77.6% vs 83.9%) and Terminal-Bench-Science (51.4% vs 59.9%).
More from Models
- Users notice Claude's latest update suddenly got more ethically preachy — debrisaway · 2026-11-13
- Baidu's ERNIE 5.5 Flash solo-produces an RSI explainer video in 8 minutes, 2-5x faster than frontier — omarsar0 · 2026-11-13
- Model Speed Depends Heavily on the Task: Wall-Clock Time Across Benchmarks — rickasaurus · 2026-11-13
- Nebius Guide: Four Ways to Adapt Open-Weight Models — SFT, Preference Tuning, RL, Distillation — demian_ai · 2026-11-13
- Claude's web fetch quietly routes pages through a small model before answering — -DankFire · 2026-11-13
- Ling-3.2-flash Lets You Toggle Thinking Mode Per Request, Docs Confirm — siddhantfuture · 2026-11-13