传 Gemini 4 Carbon 内部测试编程能力对标 Opus 5.5

Priy @alwayspriyesh原推发布于 2026.10.19

在 X 上打开原推

Google might be cooking a coding model that can compete with Claude Opus 5.5 Gemini 4 Carbon is reportedly being tested internally, and here's what we know: - An internal checkpoint reportedly tested on Google's Jetski coding platform - Early claims suggest coding performance close to Claude Opus 5.5 - It might not launch as a separate model - Carbon could eventually become an upgrade to Gemini 4 Argon Nothing is officially confirmed yet Is Google preparing a serious comeback in AI coding?