Google launches Gemini 3.8 Flash to boost vibe coding
Google’s new AI model, Gemini 3.8 Flash, may outperform competitors in coding tasks and boost its "vibe coding" push, with 75% of Google’s code now AI-generated. This could reshape its position in th…
Google is set to release Gemini 3.8 Flash as soon as today, a new AI model from DeepMind that reportedly outperforms Anthropic’s Opus in coding tasks, according to internal tests cited by the Wall Street Journal.
The move comes as Google doubles down on "vibe coding"—AI-assisted programming—after CEO Sundar Pichai revealed in March that 75% of Google’s own code is now AI-generated and engineer-approved. The push underscores the growing competition in AI coding tools, with rivals like Claude Cowork, Cursor, and OpenAI’s Codex already established in the space. Google’s rapid iteration, including the recent launch of Gemini 3.7 Flash, signals its ambition to close the gap.
Gemini 3.8 Flash arrives amid a reshuffle at DeepMind, where co-founder Demis Hassabis has taken a broader role overseeing Alphabet’s AI development. The model’s release also follows the buzz around z.AI’s GLM-5.3-Flash, which gained attention as a stealth contender on OpenRouter. While Google’s current Gemini 3.7 Flash ranks 17th on the vibe coding leaderboard, the new version aims to leapfrog competitors like Anthropic’s Claude Fable and OpenAI’s GPT-5.6 Sol, which dominate the top 10.
If the model delivers on its promise, it could redefine Google’s position in AI-assisted coding, a field critical for software development and enterprise adoption. The company’s decision to skip Gemini 3.5 Pro—due to lackluster improvements—suggests a focus on more significant upgrades, possibly paving the way for Gemini 4.0 Pro in the future. For now, developers will be watching closely to see if 3.8 Flash lives up to the hype.
Read Full Story at Android Authority →


