Google Releases Gemini 3.8 Flash

Google released Gemini 3.8 Flash. Artificial Analysis reports that the model scores 59 on its Intelligence Index and reaches the Intelligence vs. Cost per Task Pareto frontier. On the AA-Briefcase agentic evaluation the model records an Elo of 1213. With high reasoning it averages 48k output tokens per task, a 30 percent increase over 3.7 Flash. With low reasoning it averages 14k tokens. Philipp Schmid of Google DeepMind states that 3.8 is both a lot better than 3.7 and blazingly fast.