Google debuts Gemini 3.7 flash improving coding and web development
Google just dropped Gemini 3.7 Flash, its latest AI upgrade, only three weeks after the previous version.
This new model is all about making coding, web development, and knowledge tasks smoother, thanks to feedback from developers and smarter algorithms.
The numbers show big jumps: DeepSWE v1.1 scores went from 49.0% to 65.3%, and FrontierCode 1.1 Main performance climbed from 34.4% to 43.6%.
Gemini 3.7 flash benchmarks, safeguards, pricing
Gemini 3.7 Flash now creates better web designs with fewer prompts, boosting its Elo score on WebDev Arena from 1538 to 1588.
It also nails knowledge-based tasks with a GDP.pdf benchmark accuracy bump (now at 34.0%, up from 22.0%) and improved AutomationBench results (30.4% vs. 17.0%).
Security has been tightened with updated safeguards against misuse in the domains of Chemical, Biological, Radiological, and Nuclear (CBRN) and cyber offense, while enabling beneficial use cases,
and you can use Gemini on Spark (AI Pro and Ultra subscription required), Google Antigravity, or AI Studio, with pricing set at $0.75 per million input tokens and $3.75 per million output tokens through the end of 2026.