Every's week-long Gemini 4 vibe check: team splits 3-3 on writing, coding, and design
every · x · 2026-10-14
- Six Every team members spent a week testing Google DeepMind's new Gemini 4 Argon across writing, coding, design, and knowledge work — splitting 3 to 3: three psyched, three okay-but-wouldn't-use-daily.
- The appeal: sharp feedback on essay structure and metaphors, fast low-effort prototypes, and ambitious designs steerable with a clear brief. DeepMind reportedly hired professional writers to stamp out AI slop, and it shows.
- The reservations: dense one-shot drafts, overbuilt interfaces, and an app that still feels behind competitors.
- Verdict: a breakthrough if you live in Google's ecosystem, a strong secondary model for frontier hoppers — excitement doesn't mean replacing your daily driver.
More from Models
- OncologyBench humbles frontier AI: best model Claude Opus 5 scores just 40% on cancer tasks — kenbwork · 2026-10-14
- Interconnects launches tracked-models repo, explains why HuggingFace downloads are shifting — natolambert · 2026-10-14
- ChatGPT downweights self-promotional listicles, goes straight to Reddit for reviews — lilyraynyc · 2026-10-14
- First full Gemini 4 Argon hands-on review lands on YouTube: "Gemini is back" — Away_Cold_4943 · 2026-10-14
- Gemini 4 catches most fake API methods in hallucination stress test, one miss remains — haider1 · 2026-10-14
- Bindu Reddy predicts Fable 5.5 and Astra 6.1 launches within two weeks, with price cuts — bindureddy · 2026-10-14