Google Gemini 3.7 Flash Model Released with Improved Performance

Google has released Gemini 3.7 Flash, its latest model in the Gemini series, just three weeks after the previous one. This accelerated cadence is a direct result of developer feedback and algorithmic innovations that Google aims to bring to future models.

The new model boasts substantial improvements across software engineering, web development, and knowledge work. In coding, it shows strong gains over 3.6 Flash for debugging and issue resolution, with the DeepSWE v1.1 benchmark increasing from 49.0% to 65.3%. It also improves on FrontierCode 1.1 Main, going from 34.4% to 43.6%.

In web development, Gemini 3.7 Flash generates more functional layouts and feature-complete apps in fewer prompts, with an Elo score of 1588 on Arena.ai's WebDev Arena compared to the previous model's 1538. For UI generation, it shows high design adherence and parity based on a reference input.

In finance, law, biosciences, and other knowledge-dense fields, there are improvements in reasoning and accuracy, with Gemini 3.7 Flash significantly outperforming 3.6 Flash on the GDP.pdf benchmark (34.0% vs 22.0%) and AutomationBench.

The model has also updated safeguards against misuse in the domains of Chemical, Biological, Radiological, and Nuclear (CBRN) and cyber offense, while enabling beneficial use cases.