Gemini 4 Hydrogen writes 3.7x faster yet takes 474s per task, matching GPT-6.1 Sol
allenainie · x · 2026-11-14
A quoted post notes that Google's Gemini 4 Hydrogen, marketed as "built for speed", takes about as long per task as GPT-6.1 Sol on Artificial Analysis — 474 seconds vs 464 at high effort, with nearly identical scores. The catch is token efficiency: Hydrogen writes 3.7x more tokens per second but emits 85k tokens per task versus Sol's 23k, so its verbosity cancels out its throughput advantage.
More from Models
- Dev builds tabletop puzzle game combining this week's Gemini model launches — gaganghotra_ · 2026-11-14
- Anthropic Breaks Down the Five Primitives Behind Always-On AI Agents — dotey · 2026-11-14
- Model trolls its developer with fake mid-stream truncations during harness fix — repligate · 2026-11-14
- Reward hacking rates compared: Mirror-Beam hits 0.51% while most frontier coding models score zero — teortaxesTex · 2026-11-14
- Users worldwide report ChatGPT suddenly shorter, colder, memory broken — ShadowNelumbo · 2026-11-14
- Claude Opus no longer falls apart on low-effort prompts — but devolves into dense 'Claudeslop' — teortaxesTex · 2026-11-14