Gemini 3.7 Flash launches three weeks after last model, live in Spark

Google has launched Gemini 3.7 Flash, a refined AI model focusing on coding and agent workflows. Released just three weeks after Gemini 3.6 Flash, it features algorithmic reasoning improvements and a 1M-token context window. The model shows significant gains in FrontierCode 1.1 and AutomationBench, ranking 20th in Agent Arena. Available for API and enterprise users, it offers introductory pricing of $0.75 per 1M input tokens and $3.75 per 1M output tokens through December 31.