In defense of slow Gemini: async coding favors strong models over fast-but-dumb ones
Aizkmusic · x · 2026-11-12
The author pushes back on criticism that Gemini Argon is slow. Compared to the earlier era of extremely fast/cheap/dumb LLMs, slow-but-strong is a welcome shift — models are tools you need to size and match to the task (hence Jevons). It's unlike Google's usual speed-first approach, but he reads it as Google embracing async coding as a genuinely new paradigm. Given Google's scale, the earlier fast-but-dumb bet was understandable.
More from coding & agent
- Alchemy author demos Z3-backed Prove/Explore for TypeScript domain invariants — samgoodwin89 · 2026-11-12
- SteerCamp user racks up 1,182 agent memory items in 2 weeks with natural recall — pramodk73 · 2026-11-12
- 3 GitHub repos worth reading for the code: nanoGPT, tinygrad, llama.cpp — cneuralnetwork · 2026-11-12
- Reddit comparison of 7 agent messaging approaches: A2A, email, Slack, custom protocols — StrongCoffee85 · 2026-11-12
- Dev feeds new Claude a Call of Duty video, ships a playable browser tactical shooter — TAbrodi · 2026-11-12
- Second V8 patch lands: TypedArray indexOf/includes up to ~30x faster on Uint8Array — cnakazawa · 2026-11-12